This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
utf8.c: Stop using two functions
[perl5.git] / toke.c
CommitLineData
a0d0e21e 1/* toke.c
a687059c 2 *
1129b882
NC
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
a687059c 5 *
d48672a2
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
378cc40b 8 *
a0d0e21e
LW
9 */
10
11/*
4ac71550
TC
12 * 'It all comes from here, the stench and the peril.' --Frodo
13 *
14 * [p.719 of _The Lord of the Rings_, IV/ix: "Shelob's Lair"]
378cc40b
LW
15 */
16
9cbb5ea2
GS
17/*
18 * This file is the lexer for Perl. It's closely linked to the
4e553d73 19 * parser, perly.y.
ffb4593c
NT
20 *
21 * The main routine is yylex(), which returns the next token.
22 */
23
f0e67a1d
Z
24/*
25=head1 Lexer interface
26
27This is the lower layer of the Perl parser, managing characters and tokens.
28
29=for apidoc AmU|yy_parser *|PL_parser
30
31Pointer to a structure encapsulating the state of the parsing operation
32currently in progress. The pointer can be locally changed to perform
33a nested parse without interfering with the state of an outer parse.
34Individual members of C<PL_parser> have their own documentation.
35
36=cut
37*/
38
378cc40b 39#include "EXTERN.h"
864dbfa3 40#define PERL_IN_TOKE_C
378cc40b 41#include "perl.h"
04e98a4d 42#include "dquote_static.c"
378cc40b 43
eb0d8d16
NC
44#define new_constant(a,b,c,d,e,f,g) \
45 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
46
6154021b 47#define pl_yylval (PL_parser->yylval)
d3b6f988 48
199e78b7
DM
49/* XXX temporary backwards compatibility */
50#define PL_lex_brackets (PL_parser->lex_brackets)
78cdf107
Z
51#define PL_lex_allbrackets (PL_parser->lex_allbrackets)
52#define PL_lex_fakeeof (PL_parser->lex_fakeeof)
199e78b7
DM
53#define PL_lex_brackstack (PL_parser->lex_brackstack)
54#define PL_lex_casemods (PL_parser->lex_casemods)
55#define PL_lex_casestack (PL_parser->lex_casestack)
56#define PL_lex_defer (PL_parser->lex_defer)
57#define PL_lex_dojoin (PL_parser->lex_dojoin)
58#define PL_lex_expect (PL_parser->lex_expect)
59#define PL_lex_formbrack (PL_parser->lex_formbrack)
60#define PL_lex_inpat (PL_parser->lex_inpat)
61#define PL_lex_inwhat (PL_parser->lex_inwhat)
62#define PL_lex_op (PL_parser->lex_op)
63#define PL_lex_repl (PL_parser->lex_repl)
64#define PL_lex_starts (PL_parser->lex_starts)
65#define PL_lex_stuff (PL_parser->lex_stuff)
66#define PL_multi_start (PL_parser->multi_start)
67#define PL_multi_open (PL_parser->multi_open)
68#define PL_multi_close (PL_parser->multi_close)
199e78b7
DM
69#define PL_preambled (PL_parser->preambled)
70#define PL_sublex_info (PL_parser->sublex_info)
bdc0bf6f 71#define PL_linestr (PL_parser->linestr)
c2598295
DM
72#define PL_expect (PL_parser->expect)
73#define PL_copline (PL_parser->copline)
f06b5848
DM
74#define PL_bufptr (PL_parser->bufptr)
75#define PL_oldbufptr (PL_parser->oldbufptr)
76#define PL_oldoldbufptr (PL_parser->oldoldbufptr)
77#define PL_linestart (PL_parser->linestart)
78#define PL_bufend (PL_parser->bufend)
79#define PL_last_uni (PL_parser->last_uni)
80#define PL_last_lop (PL_parser->last_lop)
81#define PL_last_lop_op (PL_parser->last_lop_op)
bc177e6b 82#define PL_lex_state (PL_parser->lex_state)
2f9285f8 83#define PL_rsfp (PL_parser->rsfp)
5486870f 84#define PL_rsfp_filters (PL_parser->rsfp_filters)
12bd6ede
DM
85#define PL_in_my (PL_parser->in_my)
86#define PL_in_my_stash (PL_parser->in_my_stash)
14047fc9 87#define PL_tokenbuf (PL_parser->tokenbuf)
670a9cb2 88#define PL_multi_end (PL_parser->multi_end)
13765c85 89#define PL_error_count (PL_parser->error_count)
199e78b7
DM
90
91#ifdef PERL_MAD
92# define PL_endwhite (PL_parser->endwhite)
93# define PL_faketokens (PL_parser->faketokens)
94# define PL_lasttoke (PL_parser->lasttoke)
95# define PL_nextwhite (PL_parser->nextwhite)
96# define PL_realtokenstart (PL_parser->realtokenstart)
97# define PL_skipwhite (PL_parser->skipwhite)
98# define PL_thisclose (PL_parser->thisclose)
99# define PL_thismad (PL_parser->thismad)
100# define PL_thisopen (PL_parser->thisopen)
101# define PL_thisstuff (PL_parser->thisstuff)
102# define PL_thistoken (PL_parser->thistoken)
103# define PL_thiswhite (PL_parser->thiswhite)
fb205e7a
DM
104# define PL_thiswhite (PL_parser->thiswhite)
105# define PL_nexttoke (PL_parser->nexttoke)
106# define PL_curforce (PL_parser->curforce)
107#else
108# define PL_nexttoke (PL_parser->nexttoke)
109# define PL_nexttype (PL_parser->nexttype)
110# define PL_nextval (PL_parser->nextval)
199e78b7
DM
111#endif
112
a1894d81 113static const char* const ident_too_long = "Identifier too long";
8903cb82 114
29595ff2 115#ifdef PERL_MAD
29595ff2 116# define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
cd81e915 117# define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
9ded7720 118#else
5db06880 119# define CURMAD(slot,sv)
9ded7720 120# define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
29595ff2
NC
121#endif
122
a7aaec61
Z
123#define XENUMMASK 0x3f
124#define XFAKEEOF 0x40
125#define XFAKEBRACK 0x80
9059aa12 126
39e02b42
JH
127#ifdef USE_UTF8_SCRIPTS
128# define UTF (!IN_BYTES)
2b9d42f0 129#else
802a15e9 130# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || ( !(PL_parser->lex_flags & LEX_IGNORE_UTF8_HINTS) && (PL_hints & HINT_UTF8)))
2b9d42f0 131#endif
a0ed51b3 132
b1fc3636
CJ
133/* The maximum number of characters preceding the unrecognized one to display */
134#define UNRECOGNIZED_PRECEDE_COUNT 10
135
61f0cdd9 136/* In variables named $^X, these are the legal values for X.
2b92dfce
GS
137 * 1999-02-27 mjd-perl-patch@plover.com */
138#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
139
14bd96d0 140#define SPACE_OR_TAB(c) isBLANK_A(c)
bf4acbe4 141
ffb4593c
NT
142/* LEX_* are values for PL_lex_state, the state of the lexer.
143 * They are arranged oddly so that the guard on the switch statement
79072805 144 * can get by with a single comparison (if the compiler is smart enough).
9da1dd8f
DM
145 *
146 * These values refer to the various states within a sublex parse,
147 * i.e. within a double quotish string
79072805
LW
148 */
149
fb73857a
PP
150/* #define LEX_NOTPARSING 11 is done in perl.h. */
151
b6007c36
DM
152#define LEX_NORMAL 10 /* normal code (ie not within "...") */
153#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
154#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
155#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
156#define LEX_INTERPSTART 6 /* expecting the start of a $var */
157
158 /* at end of code, eg "$x" followed by: */
159#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
160#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
161
162#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
163 string or after \E, $foo, etc */
164#define LEX_INTERPCONST 2 /* NOT USED */
165#define LEX_FORMLINE 1 /* expecting a format line */
166#define LEX_KNOWNEXT 0 /* next token known; just return it */
167
79072805 168
bbf60fe6 169#ifdef DEBUGGING
27da23d5 170static const char* const lex_state_names[] = {
bbf60fe6
DM
171 "KNOWNEXT",
172 "FORMLINE",
173 "INTERPCONST",
174 "INTERPCONCAT",
175 "INTERPENDMAYBE",
176 "INTERPEND",
177 "INTERPSTART",
178 "INTERPPUSH",
179 "INTERPCASEMOD",
180 "INTERPNORMAL",
181 "NORMAL"
182};
183#endif
184
79072805
LW
185#ifdef ff_next
186#undef ff_next
d48672a2
LW
187#endif
188
79072805 189#include "keywords.h"
fe14fcc3 190
ffb4593c
NT
191/* CLINE is a macro that ensures PL_copline has a sane value */
192
ae986130
LW
193#ifdef CLINE
194#undef CLINE
195#endif
57843af0 196#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
3280af22 197
5db06880 198#ifdef PERL_MAD
29595ff2
NC
199# define SKIPSPACE0(s) skipspace0(s)
200# define SKIPSPACE1(s) skipspace1(s)
201# define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
202# define PEEKSPACE(s) skipspace2(s,0)
203#else
204# define SKIPSPACE0(s) skipspace(s)
205# define SKIPSPACE1(s) skipspace(s)
206# define SKIPSPACE2(s,tsv) skipspace(s)
207# define PEEKSPACE(s) skipspace(s)
208#endif
209
ffb4593c
NT
210/*
211 * Convenience functions to return different tokens and prime the
9cbb5ea2 212 * lexer for the next token. They all take an argument.
ffb4593c
NT
213 *
214 * TOKEN : generic token (used for '(', DOLSHARP, etc)
215 * OPERATOR : generic operator
216 * AOPERATOR : assignment operator
217 * PREBLOCK : beginning the block after an if, while, foreach, ...
218 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
219 * PREREF : *EXPR where EXPR is not a simple identifier
220 * TERM : expression term
221 * LOOPX : loop exiting command (goto, last, dump, etc)
222 * FTST : file test operator
223 * FUN0 : zero-argument function
7eb971ee 224 * FUN0OP : zero-argument function, with its op created in this file
2d2e263d 225 * FUN1 : not used, except for not, which isn't a UNIOP
ffb4593c
NT
226 * BOop : bitwise or or xor
227 * BAop : bitwise and
228 * SHop : shift operator
229 * PWop : power operator
9cbb5ea2 230 * PMop : pattern-matching operator
ffb4593c
NT
231 * Aop : addition-level operator
232 * Mop : multiplication-level operator
233 * Eop : equality-testing operator
e5edeb50 234 * Rop : relational operator <= != gt
ffb4593c
NT
235 *
236 * Also see LOP and lop() below.
237 */
238
998054bd 239#ifdef DEBUGGING /* Serve -DT. */
704d4215 240# define REPORT(retval) tokereport((I32)retval, &pl_yylval)
998054bd 241#else
bbf60fe6 242# define REPORT(retval) (retval)
998054bd
SC
243#endif
244
bbf60fe6
DM
245#define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
246#define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
247#define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
248#define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
249#define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
250#define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
251#define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
6154021b
RGS
252#define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
253#define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
254#define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
7eb971ee 255#define FUN0OP(f) return (pl_yylval.opval=f, CLINE, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0OP))
6154021b
RGS
256#define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
257#define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
258#define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
259#define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
260#define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
261#define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
262#define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
263#define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
264#define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
265#define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
2f3197b3 266
a687059c
LW
267/* This bit of chicanery makes a unary function followed by
268 * a parenthesis into a function with one argument, highest precedence.
6f33ba73
RGS
269 * The UNIDOR macro is for unary functions that can be followed by the //
270 * operator (such as C<shift // 0>).
a687059c 271 */
d68ce4ac 272#define UNI3(f,x,have_x) { \
6154021b 273 pl_yylval.ival = f; \
d68ce4ac 274 if (have_x) PL_expect = x; \
376fcdbf
AL
275 PL_bufptr = s; \
276 PL_last_uni = PL_oldbufptr; \
277 PL_last_lop_op = f; \
278 if (*s == '(') \
279 return REPORT( (int)FUNC1 ); \
29595ff2 280 s = PEEKSPACE(s); \
376fcdbf
AL
281 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
282 }
d68ce4ac
FC
283#define UNI(f) UNI3(f,XTERM,1)
284#define UNIDOR(f) UNI3(f,XTERMORDORDOR,1)
b5fb7ce3
FC
285#define UNIPROTO(f,optional) { \
286 if (optional) PL_last_uni = PL_oldbufptr; \
22393538
MH
287 OPERATOR(f); \
288 }
a687059c 289
d68ce4ac 290#define UNIBRACK(f) UNI3(f,0,0)
79072805 291
9f68db38 292/* grandfather return to old style */
78cdf107
Z
293#define OLDLOP(f) \
294 do { \
295 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC) \
296 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC; \
297 pl_yylval.ival = (f); \
298 PL_expect = XTERM; \
299 PL_bufptr = s; \
300 return (int)LSTOP; \
301 } while(0)
79072805 302
83944c01
FC
303#define COPLINE_INC_WITH_HERELINES \
304 STMT_START { \
305 CopLINE_inc(PL_curcop); \
d794b522
FC
306 if (PL_parser->lex_shared->herelines) \
307 CopLINE(PL_curcop) += PL_parser->lex_shared->herelines, \
308 PL_parser->lex_shared->herelines = 0; \
83944c01
FC
309 } STMT_END
310
311
8fa7f367
JH
312#ifdef DEBUGGING
313
6154021b 314/* how to interpret the pl_yylval associated with the token */
bbf60fe6
DM
315enum token_type {
316 TOKENTYPE_NONE,
317 TOKENTYPE_IVAL,
6154021b 318 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
bbf60fe6 319 TOKENTYPE_PVAL,
aeaef349 320 TOKENTYPE_OPVAL
bbf60fe6
DM
321};
322
6d4a66ac
NC
323static struct debug_tokens {
324 const int token;
325 enum token_type type;
326 const char *name;
327} const debug_tokens[] =
9041c2e3 328{
bbf60fe6
DM
329 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
330 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
331 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
332 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
333 { ARROW, TOKENTYPE_NONE, "ARROW" },
334 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
335 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
336 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
337 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
338 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
0d863452 339 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
bbf60fe6
DM
340 { DO, TOKENTYPE_NONE, "DO" },
341 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
342 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
343 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
344 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
345 { ELSE, TOKENTYPE_NONE, "ELSE" },
346 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
347 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
348 { FOR, TOKENTYPE_IVAL, "FOR" },
349 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
705fe0e5
FC
350 { FORMLBRACK, TOKENTYPE_NONE, "FORMLBRACK" },
351 { FORMRBRACK, TOKENTYPE_NONE, "FORMRBRACK" },
bbf60fe6
DM
352 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
353 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
7eb971ee 354 { FUNC0OP, TOKENTYPE_OPVAL, "FUNC0OP" },
bbf60fe6
DM
355 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
356 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
357 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
0d863452 358 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
bbf60fe6
DM
359 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
360 { IF, TOKENTYPE_IVAL, "IF" },
5504e6cf 361 { LABEL, TOKENTYPE_PVAL, "LABEL" },
bbf60fe6
DM
362 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
363 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
364 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
365 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
366 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
367 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
368 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
369 { MY, TOKENTYPE_IVAL, "MY" },
bbf60fe6
DM
370 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
371 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
372 { OROP, TOKENTYPE_IVAL, "OROP" },
373 { OROR, TOKENTYPE_NONE, "OROR" },
374 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
f3f204dc 375 { PEG, TOKENTYPE_NONE, "PEG" },
88e1f1a2
JV
376 { PLUGEXPR, TOKENTYPE_OPVAL, "PLUGEXPR" },
377 { PLUGSTMT, TOKENTYPE_OPVAL, "PLUGSTMT" },
bbf60fe6
DM
378 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
379 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
380 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
381 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
382 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
383 { PREINC, TOKENTYPE_NONE, "PREINC" },
384 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
f3f204dc 385 { QWLIST, TOKENTYPE_OPVAL, "QWLIST" },
bbf60fe6
DM
386 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
387 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
f3f204dc 388 { REQUIRE, TOKENTYPE_NONE, "REQUIRE" },
bbf60fe6
DM
389 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
390 { SUB, TOKENTYPE_NONE, "SUB" },
391 { THING, TOKENTYPE_OPVAL, "THING" },
392 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
393 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
394 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
395 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
396 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
397 { USE, TOKENTYPE_IVAL, "USE" },
0d863452 398 { WHEN, TOKENTYPE_IVAL, "WHEN" },
bbf60fe6
DM
399 { WHILE, TOKENTYPE_IVAL, "WHILE" },
400 { WORD, TOKENTYPE_OPVAL, "WORD" },
be25f609 401 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
c35e046a 402 { 0, TOKENTYPE_NONE, NULL }
bbf60fe6
DM
403};
404
6154021b 405/* dump the returned token in rv, plus any optional arg in pl_yylval */
998054bd 406
bbf60fe6 407STATIC int
704d4215 408S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
bbf60fe6 409{
97aff369 410 dVAR;
7918f24d
NC
411
412 PERL_ARGS_ASSERT_TOKEREPORT;
413
bbf60fe6 414 if (DEBUG_T_TEST) {
bd61b366 415 const char *name = NULL;
bbf60fe6 416 enum token_type type = TOKENTYPE_NONE;
f54cb97a 417 const struct debug_tokens *p;
396482e1 418 SV* const report = newSVpvs("<== ");
bbf60fe6 419
f54cb97a 420 for (p = debug_tokens; p->token; p++) {
bbf60fe6
DM
421 if (p->token == (int)rv) {
422 name = p->name;
423 type = p->type;
424 break;
425 }
426 }
427 if (name)
54667de8 428 Perl_sv_catpv(aTHX_ report, name);
74736ae6 429 else if ((char)rv > ' ' && (char)rv <= '~')
4ebc7986 430 {
bbf60fe6 431 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
4ebc7986
FC
432 if ((char)rv == 'p')
433 sv_catpvs(report, " (pending identifier)");
434 }
bbf60fe6 435 else if (!rv)
396482e1 436 sv_catpvs(report, "EOF");
bbf60fe6
DM
437 else
438 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
439 switch (type) {
440 case TOKENTYPE_NONE:
bbf60fe6
DM
441 break;
442 case TOKENTYPE_IVAL:
704d4215 443 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
bbf60fe6
DM
444 break;
445 case TOKENTYPE_OPNUM:
446 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
704d4215 447 PL_op_name[lvalp->ival]);
bbf60fe6
DM
448 break;
449 case TOKENTYPE_PVAL:
704d4215 450 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
bbf60fe6
DM
451 break;
452 case TOKENTYPE_OPVAL:
704d4215 453 if (lvalp->opval) {
401441c0 454 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
704d4215
GG
455 PL_op_name[lvalp->opval->op_type]);
456 if (lvalp->opval->op_type == OP_CONST) {
b6007c36 457 Perl_sv_catpvf(aTHX_ report, " %s",
704d4215 458 SvPEEK(cSVOPx_sv(lvalp->opval)));
b6007c36
DM
459 }
460
461 }
401441c0 462 else
396482e1 463 sv_catpvs(report, "(opval=null)");
bbf60fe6
DM
464 break;
465 }
b6007c36 466 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
bbf60fe6
DM
467 };
468 return (int)rv;
998054bd
SC
469}
470
b6007c36
DM
471
472/* print the buffer with suitable escapes */
473
474STATIC void
15f169a1 475S_printbuf(pTHX_ const char *const fmt, const char *const s)
b6007c36 476{
396482e1 477 SV* const tmp = newSVpvs("");
7918f24d
NC
478
479 PERL_ARGS_ASSERT_PRINTBUF;
480
b6007c36
DM
481 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
482 SvREFCNT_dec(tmp);
483}
484
8fa7f367
JH
485#endif
486
8290c323
NC
487static int
488S_deprecate_commaless_var_list(pTHX) {
489 PL_expect = XTERM;
490 deprecate("comma-less variable list");
491 return REPORT(','); /* grandfather non-comma-format format */
492}
493
ffb4593c
NT
494/*
495 * S_ao
496 *
c963b151
BD
497 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
498 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
ffb4593c
NT
499 */
500
76e3520e 501STATIC int
cea2e8a9 502S_ao(pTHX_ int toketype)
a0d0e21e 503{
97aff369 504 dVAR;
3280af22
NIS
505 if (*PL_bufptr == '=') {
506 PL_bufptr++;
a0d0e21e 507 if (toketype == ANDAND)
6154021b 508 pl_yylval.ival = OP_ANDASSIGN;
a0d0e21e 509 else if (toketype == OROR)
6154021b 510 pl_yylval.ival = OP_ORASSIGN;
c963b151 511 else if (toketype == DORDOR)
6154021b 512 pl_yylval.ival = OP_DORASSIGN;
a0d0e21e
LW
513 toketype = ASSIGNOP;
514 }
515 return toketype;
516}
517
ffb4593c
NT
518/*
519 * S_no_op
520 * When Perl expects an operator and finds something else, no_op
521 * prints the warning. It always prints "<something> found where
522 * operator expected. It prints "Missing semicolon on previous line?"
523 * if the surprise occurs at the start of the line. "do you need to
524 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
525 * where the compiler doesn't know if foo is a method call or a function.
526 * It prints "Missing operator before end of line" if there's nothing
527 * after the missing operator, or "... before <...>" if there is something
528 * after the missing operator.
529 */
530
76e3520e 531STATIC void
15f169a1 532S_no_op(pTHX_ const char *const what, char *s)
463ee0b2 533{
97aff369 534 dVAR;
9d4ba2ae
AL
535 char * const oldbp = PL_bufptr;
536 const bool is_first = (PL_oldbufptr == PL_linestart);
68dc0745 537
7918f24d
NC
538 PERL_ARGS_ASSERT_NO_OP;
539
1189a94a
GS
540 if (!s)
541 s = oldbp;
07c798fb 542 else
1189a94a 543 PL_bufptr = s;
734ab321 544 yywarn(Perl_form(aTHX_ "%s found where operator expected", what), UTF ? SVf_UTF8 : 0);
56da5a46
RGS
545 if (ckWARN_d(WARN_SYNTAX)) {
546 if (is_first)
547 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
548 "\t(Missing semicolon on previous line?)\n");
549 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
f54cb97a 550 const char *t;
8a2bca12 551 for (t = PL_oldoldbufptr; (isWORDCHAR_lazy_if(t,UTF) || *t == ':');
734ab321 552 t += UTF ? UTF8SKIP(t) : 1)
c35e046a 553 NOOP;
56da5a46
RGS
554 if (t < PL_bufptr && isSPACE(*t))
555 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
b17a0679
FC
556 "\t(Do you need to predeclare %"UTF8f"?)\n",
557 UTF8fARG(UTF, t - PL_oldoldbufptr, PL_oldoldbufptr));
56da5a46
RGS
558 }
559 else {
560 assert(s >= oldbp);
561 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
b17a0679
FC
562 "\t(Missing operator before %"UTF8f"?)\n",
563 UTF8fARG(UTF, s - oldbp, oldbp));
56da5a46 564 }
07c798fb 565 }
3280af22 566 PL_bufptr = oldbp;
8990e307
LW
567}
568
ffb4593c
NT
569/*
570 * S_missingterm
571 * Complain about missing quote/regexp/heredoc terminator.
d4c19fe8 572 * If it's called with NULL then it cauterizes the line buffer.
ffb4593c
NT
573 * If we're in a delimited string and the delimiter is a control
574 * character, it's reformatted into a two-char sequence like ^C.
575 * This is fatal.
576 */
577
76e3520e 578STATIC void
cea2e8a9 579S_missingterm(pTHX_ char *s)
8990e307 580{
97aff369 581 dVAR;
8990e307
LW
582 char tmpbuf[3];
583 char q;
584 if (s) {
9d4ba2ae 585 char * const nl = strrchr(s,'\n');
d2719217 586 if (nl)
8990e307
LW
587 *nl = '\0';
588 }
463559e7 589 else if (isCNTRL(PL_multi_close)) {
8990e307 590 *tmpbuf = '^';
585ec06d 591 tmpbuf[1] = (char)toCTRL(PL_multi_close);
8990e307
LW
592 tmpbuf[2] = '\0';
593 s = tmpbuf;
594 }
595 else {
eb160463 596 *tmpbuf = (char)PL_multi_close;
8990e307
LW
597 tmpbuf[1] = '\0';
598 s = tmpbuf;
599 }
600 q = strchr(s,'"') ? '\'' : '"';
cea2e8a9 601 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
463ee0b2 602}
79072805 603
dd0ac2b9
FC
604#include "feature.h"
605
0d863452 606/*
0d863452
RH
607 * Check whether the named feature is enabled.
608 */
26ea9e12 609bool
3fff3427 610Perl_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
0d863452 611{
97aff369 612 dVAR;
4a731d7b 613 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
7918f24d
NC
614
615 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
ca4d40c4
FC
616
617 assert(CURRENT_FEATURE_BUNDLE == FEATURE_BUNDLE_CUSTOM);
7918f24d 618
26ea9e12
NC
619 if (namelen > MAX_FEATURE_LEN)
620 return FALSE;
3fff3427 621 memcpy(&he_name[8], name, namelen);
7d69d4a6 622
c8ca97b0
NC
623 return cBOOL(cop_hints_fetch_pvn(PL_curcop, he_name, 8 + namelen, 0,
624 REFCOUNTED_HE_EXISTS));
0d863452
RH
625}
626
ffb4593c 627/*
9cbb5ea2
GS
628 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
629 * utf16-to-utf8-reversed.
ffb4593c
NT
630 */
631
c39cd008
GS
632#ifdef PERL_CR_FILTER
633static void
634strip_return(SV *sv)
635{
eb578fdb
KW
636 const char *s = SvPVX_const(sv);
637 const char * const e = s + SvCUR(sv);
7918f24d
NC
638
639 PERL_ARGS_ASSERT_STRIP_RETURN;
640
c39cd008
GS
641 /* outer loop optimized to do nothing if there are no CR-LFs */
642 while (s < e) {
643 if (*s++ == '\r' && *s == '\n') {
644 /* hit a CR-LF, need to copy the rest */
eb578fdb 645 char *d = s - 1;
c39cd008
GS
646 *d++ = *s++;
647 while (s < e) {
648 if (*s == '\r' && s[1] == '\n')
649 s++;
650 *d++ = *s++;
651 }
652 SvCUR(sv) -= s - d;
653 return;
654 }
655 }
656}
a868473f 657
76e3520e 658STATIC I32
c39cd008 659S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
a868473f 660{
f54cb97a 661 const I32 count = FILTER_READ(idx+1, sv, maxlen);
c39cd008
GS
662 if (count > 0 && !maxlen)
663 strip_return(sv);
664 return count;
a868473f
NIS
665}
666#endif
667
ffb4593c 668/*
8eaa0acf
Z
669=for apidoc Amx|void|lex_start|SV *line|PerlIO *rsfp|U32 flags
670
671Creates and initialises a new lexer/parser state object, supplying
672a context in which to lex and parse from a new source of Perl code.
673A pointer to the new state object is placed in L</PL_parser>. An entry
674is made on the save stack so that upon unwinding the new state object
675will be destroyed and the former value of L</PL_parser> will be restored.
676Nothing else need be done to clean up the parsing context.
677
678The code to be parsed comes from I<line> and I<rsfp>. I<line>, if
679non-null, provides a string (in SV form) containing code to be parsed.
680A copy of the string is made, so subsequent modification of I<line>
681does not affect parsing. I<rsfp>, if non-null, provides an input stream
682from which code will be read to be parsed. If both are non-null, the
683code in I<line> comes first and must consist of complete lines of input,
684and I<rsfp> supplies the remainder of the source.
685
e368b3bd
FC
686The I<flags> parameter is reserved for future use. Currently it is only
687used by perl internally, so extensions should always pass zero.
8eaa0acf
Z
688
689=cut
690*/
ffb4593c 691
27fcb6ee 692/* LEX_START_SAME_FILTER indicates that this is not a new file, so it
87606032
NC
693 can share filters with the current parser.
694 LEX_START_DONT_CLOSE indicates that the file handle wasn't opened by the
695 caller, hence isn't owned by the parser, so shouldn't be closed on parser
696 destruction. This is used to handle the case of defaulting to reading the
697 script from the standard input because no filename was given on the command
698 line (without getting confused by situation where STDIN has been closed, so
699 the script handle is opened on fd 0) */
27fcb6ee 700
a0d0e21e 701void
8eaa0acf 702Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, U32 flags)
79072805 703{
97aff369 704 dVAR;
6ef55633 705 const char *s = NULL;
5486870f 706 yy_parser *parser, *oparser;
60d63348 707 if (flags && flags & ~LEX_START_FLAGS)
8eaa0acf 708 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_start");
acdf0a21
DM
709
710 /* create and initialise a parser */
711
199e78b7 712 Newxz(parser, 1, yy_parser);
5486870f 713 parser->old_parser = oparser = PL_parser;
acdf0a21
DM
714 PL_parser = parser;
715
28ac2b49
Z
716 parser->stack = NULL;
717 parser->ps = NULL;
718 parser->stack_size = 0;
acdf0a21 719
e3abe207
DM
720 /* on scope exit, free this parser and restore any outer one */
721 SAVEPARSER(parser);
7c4baf47 722 parser->saved_curcop = PL_curcop;
e3abe207 723
acdf0a21 724 /* initialise lexer state */
8990e307 725
fb205e7a
DM
726#ifdef PERL_MAD
727 parser->curforce = -1;
728#else
729 parser->nexttoke = 0;
730#endif
ca4cfd28 731 parser->error_count = oparser ? oparser->error_count : 0;
c2598295 732 parser->copline = NOLINE;
5afb0a62 733 parser->lex_state = LEX_NORMAL;
c2598295 734 parser->expect = XSTATE;
2f9285f8 735 parser->rsfp = rsfp;
27fcb6ee
FC
736 parser->rsfp_filters =
737 !(flags & LEX_START_SAME_FILTER) || !oparser
d3cd8e11
FC
738 ? NULL
739 : MUTABLE_AV(SvREFCNT_inc(
740 oparser->rsfp_filters
741 ? oparser->rsfp_filters
742 : (oparser->rsfp_filters = newAV())
743 ));
2f9285f8 744
199e78b7
DM
745 Newx(parser->lex_brackstack, 120, char);
746 Newx(parser->lex_casestack, 12, char);
747 *parser->lex_casestack = '\0';
d794b522 748 Newxz(parser->lex_shared, 1, LEXSHARED);
02b34bbe 749
10efb74f 750 if (line) {
0528fd32 751 STRLEN len;
10efb74f 752 s = SvPV_const(line, len);
0abcdfa4
FC
753 parser->linestr = flags & LEX_START_COPIED
754 ? SvREFCNT_inc_simple_NN(line)
755 : newSVpvn_flags(s, len, SvUTF8(line));
bf1b738b 756 sv_catpvn(parser->linestr, "\n;", rsfp ? 1 : 2);
0abcdfa4 757 } else {
bf1b738b 758 parser->linestr = newSVpvn("\n;", rsfp ? 1 : 2);
8990e307 759 }
f06b5848
DM
760 parser->oldoldbufptr =
761 parser->oldbufptr =
762 parser->bufptr =
763 parser->linestart = SvPVX(parser->linestr);
764 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
765 parser->last_lop = parser->last_uni = NULL;
87606032
NC
766 parser->lex_flags = flags & (LEX_IGNORE_UTF8_HINTS|LEX_EVALBYTES
767 |LEX_DONT_CLOSE_RSFP);
737c24fc 768
60d63348 769 parser->in_pod = parser->filtered = 0;
79072805 770}
a687059c 771
e3abe207
DM
772
773/* delete a parser object */
774
775void
776Perl_parser_free(pTHX_ const yy_parser *parser)
777{
7918f24d
NC
778 PERL_ARGS_ASSERT_PARSER_FREE;
779
7c4baf47 780 PL_curcop = parser->saved_curcop;
bdc0bf6f
DM
781 SvREFCNT_dec(parser->linestr);
782
87606032 783 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
2f9285f8 784 PerlIO_clearerr(parser->rsfp);
799361c3
SH
785 else if (parser->rsfp && (!parser->old_parser ||
786 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
2f9285f8 787 PerlIO_close(parser->rsfp);
5486870f 788 SvREFCNT_dec(parser->rsfp_filters);
10002bc1
FC
789 SvREFCNT_dec(parser->lex_stuff);
790 SvREFCNT_dec(parser->sublex_info.repl);
3ac7ff8f
FC
791
792 Safefree(parser->lex_brackstack);
793 Safefree(parser->lex_casestack);
794 Safefree(parser->lex_shared);
795 PL_parser = parser->old_parser;
796 Safefree(parser);
797}
798
799void
800Perl_parser_free_nexttoke_ops(pTHX_ yy_parser *parser, OPSLAB *slab)
801{
802#ifdef PERL_MAD
803 I32 nexttoke = parser->lasttoke;
804#else
805 I32 nexttoke = parser->nexttoke;
806#endif
807 PERL_ARGS_ASSERT_PARSER_FREE_NEXTTOKE_OPS;
3ce3dcd9
FC
808 while (nexttoke--) {
809#ifdef PERL_MAD
810 if (S_is_opval_token(parser->nexttoke[nexttoke].next_type
3ac7ff8f
FC
811 & 0xffff)
812 && parser->nexttoke[nexttoke].next_val.opval
813 && parser->nexttoke[nexttoke].next_val.opval->op_slabbed
814 && OpSLAB(parser->nexttoke[nexttoke].next_val.opval) == slab) {
815 op_free(parser->nexttoke[nexttoke].next_val.opval);
816 parser->nexttoke[nexttoke].next_val.opval = NULL;
817 }
3ce3dcd9 818#else
3ac7ff8f
FC
819 if (S_is_opval_token(parser->nexttype[nexttoke] & 0xffff)
820 && parser->nextval[nexttoke].opval
821 && parser->nextval[nexttoke].opval->op_slabbed
822 && OpSLAB(parser->nextval[nexttoke].opval) == slab) {
3ce3dcd9 823 op_free(parser->nextval[nexttoke].opval);
3ac7ff8f
FC
824 parser->nextval[nexttoke].opval = NULL;
825 }
3ce3dcd9
FC
826#endif
827 }
e3abe207
DM
828}
829
830
ffb4593c 831/*
f0e67a1d
Z
832=for apidoc AmxU|SV *|PL_parser-E<gt>linestr
833
834Buffer scalar containing the chunk currently under consideration of the
835text currently being lexed. This is always a plain string scalar (for
836which C<SvPOK> is true). It is not intended to be used as a scalar by
837normal scalar means; instead refer to the buffer directly by the pointer
838variables described below.
839
840The lexer maintains various C<char*> pointers to things in the
841C<PL_parser-E<gt>linestr> buffer. If C<PL_parser-E<gt>linestr> is ever
842reallocated, all of these pointers must be updated. Don't attempt to
843do this manually, but rather use L</lex_grow_linestr> if you need to
844reallocate the buffer.
845
846The content of the text chunk in the buffer is commonly exactly one
847complete line of input, up to and including a newline terminator,
848but there are situations where it is otherwise. The octets of the
849buffer may be intended to be interpreted as either UTF-8 or Latin-1.
850The function L</lex_bufutf8> tells you which. Do not use the C<SvUTF8>
851flag on this scalar, which may disagree with it.
852
853For direct examination of the buffer, the variable
854L</PL_parser-E<gt>bufend> points to the end of the buffer. The current
855lexing position is pointed to by L</PL_parser-E<gt>bufptr>. Direct use
856of these pointers is usually preferable to examination of the scalar
857through normal scalar means.
858
859=for apidoc AmxU|char *|PL_parser-E<gt>bufend
860
861Direct pointer to the end of the chunk of text currently being lexed, the
862end of the lexer buffer. This is equal to C<SvPVX(PL_parser-E<gt>linestr)
863+ SvCUR(PL_parser-E<gt>linestr)>. A NUL character (zero octet) is
864always located at the end of the buffer, and does not count as part of
865the buffer's contents.
866
867=for apidoc AmxU|char *|PL_parser-E<gt>bufptr
868
869Points to the current position of lexing inside the lexer buffer.
870Characters around this point may be freely examined, within
871the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and
872L</PL_parser-E<gt>bufend>. The octets of the buffer may be intended to be
873interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>.
874
875Lexing code (whether in the Perl core or not) moves this pointer past
876the characters that it consumes. It is also expected to perform some
877bookkeeping whenever a newline character is consumed. This movement
878can be more conveniently performed by the function L</lex_read_to>,
879which handles newlines appropriately.
880
881Interpretation of the buffer's octets can be abstracted out by
882using the slightly higher-level functions L</lex_peek_unichar> and
883L</lex_read_unichar>.
884
885=for apidoc AmxU|char *|PL_parser-E<gt>linestart
886
887Points to the start of the current line inside the lexer buffer.
888This is useful for indicating at which column an error occurred, and
889not much else. This must be updated by any lexing code that consumes
890a newline; the function L</lex_read_to> handles this detail.
891
892=cut
893*/
894
895/*
896=for apidoc Amx|bool|lex_bufutf8
897
898Indicates whether the octets in the lexer buffer
899(L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding
900of Unicode characters. If not, they should be interpreted as Latin-1
901characters. This is analogous to the C<SvUTF8> flag for scalars.
902
903In UTF-8 mode, it is not guaranteed that the lexer buffer actually
904contains valid UTF-8. Lexing code must be robust in the face of invalid
905encoding.
906
907The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar
908is significant, but not the whole story regarding the input character
909encoding. Normally, when a file is being read, the scalar contains octets
910and its C<SvUTF8> flag is off, but the octets should be interpreted as
911UTF-8 if the C<use utf8> pragma is in effect. During a string eval,
912however, the scalar may have the C<SvUTF8> flag on, and in this case its
913octets should be interpreted as UTF-8 unless the C<use bytes> pragma
914is in effect. This logic may change in the future; use this function
915instead of implementing the logic yourself.
916
917=cut
918*/
919
920bool
921Perl_lex_bufutf8(pTHX)
922{
923 return UTF;
924}
925
926/*
927=for apidoc Amx|char *|lex_grow_linestr|STRLEN len
928
929Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate
930at least I<len> octets (including terminating NUL). Returns a
931pointer to the reallocated buffer. This is necessary before making
932any direct modification of the buffer that would increase its length.
933L</lex_stuff_pvn> provides a more convenient way to insert text into
934the buffer.
935
936Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>;
937this function updates all of the lexer's variables that point directly
938into the buffer.
939
940=cut
941*/
942
943char *
944Perl_lex_grow_linestr(pTHX_ STRLEN len)
945{
946 SV *linestr;
947 char *buf;
948 STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
c7641931 949 STRLEN linestart_pos, last_uni_pos, last_lop_pos, re_eval_start_pos;
f0e67a1d
Z
950 linestr = PL_parser->linestr;
951 buf = SvPVX(linestr);
952 if (len <= SvLEN(linestr))
953 return buf;
954 bufend_pos = PL_parser->bufend - buf;
955 bufptr_pos = PL_parser->bufptr - buf;
956 oldbufptr_pos = PL_parser->oldbufptr - buf;
957 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
958 linestart_pos = PL_parser->linestart - buf;
959 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
960 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
3328ab5a
FC
961 re_eval_start_pos = PL_parser->lex_shared->re_eval_start ?
962 PL_parser->lex_shared->re_eval_start - buf : 0;
c7641931 963
f0e67a1d 964 buf = sv_grow(linestr, len);
c7641931 965
f0e67a1d
Z
966 PL_parser->bufend = buf + bufend_pos;
967 PL_parser->bufptr = buf + bufptr_pos;
968 PL_parser->oldbufptr = buf + oldbufptr_pos;
969 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
970 PL_parser->linestart = buf + linestart_pos;
971 if (PL_parser->last_uni)
972 PL_parser->last_uni = buf + last_uni_pos;
973 if (PL_parser->last_lop)
974 PL_parser->last_lop = buf + last_lop_pos;
3328ab5a
FC
975 if (PL_parser->lex_shared->re_eval_start)
976 PL_parser->lex_shared->re_eval_start = buf + re_eval_start_pos;
f0e67a1d
Z
977 return buf;
978}
979
980/*
83aa740e 981=for apidoc Amx|void|lex_stuff_pvn|const char *pv|STRLEN len|U32 flags
f0e67a1d
Z
982
983Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
984immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
985reallocating the buffer if necessary. This means that lexing code that
986runs later will see the characters as if they had appeared in the input.
987It is not recommended to do this as part of normal parsing, and most
988uses of this facility run the risk of the inserted characters being
989interpreted in an unintended manner.
990
991The string to be inserted is represented by I<len> octets starting
992at I<pv>. These octets are interpreted as either UTF-8 or Latin-1,
993according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>.
994The characters are recoded for the lexer buffer, according to how the
995buffer is currently being interpreted (L</lex_bufutf8>). If a string
9dcc53ea 996to be inserted is available as a Perl scalar, the L</lex_stuff_sv>
f0e67a1d
Z
997function is more convenient.
998
999=cut
1000*/
1001
1002void
83aa740e 1003Perl_lex_stuff_pvn(pTHX_ const char *pv, STRLEN len, U32 flags)
f0e67a1d 1004{
749123ff 1005 dVAR;
f0e67a1d
Z
1006 char *bufptr;
1007 PERL_ARGS_ASSERT_LEX_STUFF_PVN;
1008 if (flags & ~(LEX_STUFF_UTF8))
1009 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn");
1010 if (UTF) {
1011 if (flags & LEX_STUFF_UTF8) {
1012 goto plain_copy;
1013 } else {
54d004e8 1014 STRLEN highhalf = 0; /* Count of variants */
83aa740e 1015 const char *p, *e = pv+len;
54d004e8
KW
1016 for (p = pv; p != e; p++) {
1017 if (! UTF8_IS_INVARIANT(*p)) {
1018 highhalf++;
1019 }
1020 }
f0e67a1d
Z
1021 if (!highhalf)
1022 goto plain_copy;
1023 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf);
1024 bufptr = PL_parser->bufptr;
1025 Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
1026 SvCUR_set(PL_parser->linestr,
1027 SvCUR(PL_parser->linestr) + len+highhalf);
f0e67a1d
Z
1028 PL_parser->bufend += len+highhalf;
1029 for (p = pv; p != e; p++) {
1030 U8 c = (U8)*p;
54d004e8
KW
1031 if (! UTF8_IS_INVARIANT(c)) {
1032 *bufptr++ = UTF8_TWO_BYTE_HI(c);
1033 *bufptr++ = UTF8_TWO_BYTE_LO(c);
f0e67a1d
Z
1034 } else {
1035 *bufptr++ = (char)c;
1036 }
1037 }
1038 }
1039 } else {
1040 if (flags & LEX_STUFF_UTF8) {
1041 STRLEN highhalf = 0;
83aa740e 1042 const char *p, *e = pv+len;
f0e67a1d
Z
1043 for (p = pv; p != e; p++) {
1044 U8 c = (U8)*p;
54d004e8 1045 if (UTF8_IS_ABOVE_LATIN1(c)) {
f0e67a1d
Z
1046 Perl_croak(aTHX_ "Lexing code attempted to stuff "
1047 "non-Latin-1 character into Latin-1 input");
54d004e8 1048 } else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(p, e)) {
f0e67a1d
Z
1049 p++;
1050 highhalf++;
54d004e8 1051 } else if (! UTF8_IS_INVARIANT(c)) {
f0e67a1d
Z
1052 /* malformed UTF-8 */
1053 ENTER;
1054 SAVESPTR(PL_warnhook);
1055 PL_warnhook = PERL_WARNHOOK_FATAL;
c80e42f3 1056 utf8n_to_uvchr((U8*)p, e-p, NULL, 0);
f0e67a1d
Z
1057 LEAVE;
1058 }
1059 }
1060 if (!highhalf)
1061 goto plain_copy;
1062 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf);
1063 bufptr = PL_parser->bufptr;
1064 Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
1065 SvCUR_set(PL_parser->linestr,
1066 SvCUR(PL_parser->linestr) + len-highhalf);
f0e67a1d 1067 PL_parser->bufend += len-highhalf;
54d004e8
KW
1068 p = pv;
1069 while (p < e) {
1070 if (UTF8_IS_INVARIANT(*p)) {
1071 *bufptr++ = *p;
1072 p++;
f0e67a1d 1073 }
54d004e8
KW
1074 else {
1075 assert(p < e -1 );
94bb8c36 1076 *bufptr++ = TWO_BYTE_UTF8_TO_NATIVE(*p, *(p+1));
54d004e8
KW
1077 p += 2;
1078 }
f0e67a1d
Z
1079 }
1080 } else {
54d004e8 1081 plain_copy:
f0e67a1d
Z
1082 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len);
1083 bufptr = PL_parser->bufptr;
1084 Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char);
255fdf19 1085 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) + len);
f0e67a1d
Z
1086 PL_parser->bufend += len;
1087 Copy(pv, bufptr, len, char);
1088 }
1089 }
1090}
1091
1092/*
9dcc53ea
Z
1093=for apidoc Amx|void|lex_stuff_pv|const char *pv|U32 flags
1094
1095Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1096immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1097reallocating the buffer if necessary. This means that lexing code that
1098runs later will see the characters as if they had appeared in the input.
1099It is not recommended to do this as part of normal parsing, and most
1100uses of this facility run the risk of the inserted characters being
1101interpreted in an unintended manner.
1102
1103The string to be inserted is represented by octets starting at I<pv>
1104and continuing to the first nul. These octets are interpreted as either
1105UTF-8 or Latin-1, according to whether the C<LEX_STUFF_UTF8> flag is set
1106in I<flags>. The characters are recoded for the lexer buffer, according
1107to how the buffer is currently being interpreted (L</lex_bufutf8>).
1108If it is not convenient to nul-terminate a string to be inserted, the
1109L</lex_stuff_pvn> function is more appropriate.
1110
1111=cut
1112*/
1113
1114void
1115Perl_lex_stuff_pv(pTHX_ const char *pv, U32 flags)
1116{
1117 PERL_ARGS_ASSERT_LEX_STUFF_PV;
1118 lex_stuff_pvn(pv, strlen(pv), flags);
1119}
1120
1121/*
f0e67a1d
Z
1122=for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags
1123
1124Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1125immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1126reallocating the buffer if necessary. This means that lexing code that
1127runs later will see the characters as if they had appeared in the input.
1128It is not recommended to do this as part of normal parsing, and most
1129uses of this facility run the risk of the inserted characters being
1130interpreted in an unintended manner.
1131
1132The string to be inserted is the string value of I<sv>. The characters
1133are recoded for the lexer buffer, according to how the buffer is currently
9dcc53ea 1134being interpreted (L</lex_bufutf8>). If a string to be inserted is
f0e67a1d
Z
1135not already a Perl scalar, the L</lex_stuff_pvn> function avoids the
1136need to construct a scalar.
1137
1138=cut
1139*/
1140
1141void
1142Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags)
1143{
1144 char *pv;
1145 STRLEN len;
1146 PERL_ARGS_ASSERT_LEX_STUFF_SV;
1147 if (flags)
1148 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv");
1149 pv = SvPV(sv, len);
1150 lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0));
1151}
1152
1153/*
1154=for apidoc Amx|void|lex_unstuff|char *ptr
1155
1156Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to
1157I<ptr>. Text following I<ptr> will be moved, and the buffer shortened.
1158This hides the discarded text from any lexing code that runs later,
1159as if the text had never appeared.
1160
1161This is not the normal way to consume lexed text. For that, use
1162L</lex_read_to>.
1163
1164=cut
1165*/
1166
1167void
1168Perl_lex_unstuff(pTHX_ char *ptr)
1169{
1170 char *buf, *bufend;
1171 STRLEN unstuff_len;
1172 PERL_ARGS_ASSERT_LEX_UNSTUFF;
1173 buf = PL_parser->bufptr;
1174 if (ptr < buf)
1175 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1176 if (ptr == buf)
1177 return;
1178 bufend = PL_parser->bufend;
1179 if (ptr > bufend)
1180 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1181 unstuff_len = ptr - buf;
1182 Move(ptr, buf, bufend+1-ptr, char);
1183 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len);
1184 PL_parser->bufend = bufend - unstuff_len;
1185}
1186
1187/*
1188=for apidoc Amx|void|lex_read_to|char *ptr
1189
1190Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up
1191to I<ptr>. This advances L</PL_parser-E<gt>bufptr> to match I<ptr>,
1192performing the correct bookkeeping whenever a newline character is passed.
1193This is the normal way to consume lexed text.
1194
1195Interpretation of the buffer's octets can be abstracted out by
1196using the slightly higher-level functions L</lex_peek_unichar> and
1197L</lex_read_unichar>.
1198
1199=cut
1200*/
1201
1202void
1203Perl_lex_read_to(pTHX_ char *ptr)
1204{
1205 char *s;
1206 PERL_ARGS_ASSERT_LEX_READ_TO;
1207 s = PL_parser->bufptr;
1208 if (ptr < s || ptr > PL_parser->bufend)
1209 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to");
1210 for (; s != ptr; s++)
1211 if (*s == '\n') {
83944c01 1212 COPLINE_INC_WITH_HERELINES;
f0e67a1d
Z
1213 PL_parser->linestart = s+1;
1214 }
1215 PL_parser->bufptr = ptr;
1216}
1217
1218/*
1219=for apidoc Amx|void|lex_discard_to|char *ptr
1220
1221Discards the first part of the L</PL_parser-E<gt>linestr> buffer,
1222up to I<ptr>. The remaining content of the buffer will be moved, and
1223all pointers into the buffer updated appropriately. I<ptr> must not
1224be later in the buffer than the position of L</PL_parser-E<gt>bufptr>:
1225it is not permitted to discard text that has yet to be lexed.
1226
1227Normally it is not necessarily to do this directly, because it suffices to
1228use the implicit discarding behaviour of L</lex_next_chunk> and things
1229based on it. However, if a token stretches across multiple lines,
1f317c95 1230and the lexing code has kept multiple lines of text in the buffer for
f0e67a1d
Z
1231that purpose, then after completion of the token it would be wise to
1232explicitly discard the now-unneeded earlier lines, to avoid future
1233multi-line tokens growing the buffer without bound.
1234
1235=cut
1236*/
1237
1238void
1239Perl_lex_discard_to(pTHX_ char *ptr)
1240{
1241 char *buf;
1242 STRLEN discard_len;
1243 PERL_ARGS_ASSERT_LEX_DISCARD_TO;
1244 buf = SvPVX(PL_parser->linestr);
1245 if (ptr < buf)
1246 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1247 if (ptr == buf)
1248 return;
1249 if (ptr > PL_parser->bufptr)
1250 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1251 discard_len = ptr - buf;
1252 if (PL_parser->oldbufptr < ptr)
1253 PL_parser->oldbufptr = ptr;
1254 if (PL_parser->oldoldbufptr < ptr)
1255 PL_parser->oldoldbufptr = ptr;
1256 if (PL_parser->last_uni && PL_parser->last_uni < ptr)
1257 PL_parser->last_uni = NULL;
1258 if (PL_parser->last_lop && PL_parser->last_lop < ptr)
1259 PL_parser->last_lop = NULL;
1260 Move(ptr, buf, PL_parser->bufend+1-ptr, char);
1261 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len);
1262 PL_parser->bufend -= discard_len;
1263 PL_parser->bufptr -= discard_len;
1264 PL_parser->oldbufptr -= discard_len;
1265 PL_parser->oldoldbufptr -= discard_len;
1266 if (PL_parser->last_uni)
1267 PL_parser->last_uni -= discard_len;
1268 if (PL_parser->last_lop)
1269 PL_parser->last_lop -= discard_len;
1270}
1271
1272/*
1273=for apidoc Amx|bool|lex_next_chunk|U32 flags
1274
1275Reads in the next chunk of text to be lexed, appending it to
1276L</PL_parser-E<gt>linestr>. This should be called when lexing code has
1277looked to the end of the current chunk and wants to know more. It is
1278usual, but not necessary, for lexing to have consumed the entirety of
1279the current chunk at this time.
1280
1281If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current
1282chunk (i.e., the current chunk has been entirely consumed), normally the
1283current chunk will be discarded at the same time that the new chunk is
1284read in. If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk
1285will not be discarded. If the current chunk has not been entirely
1286consumed, then it will not be discarded regardless of the flag.
1287
1288Returns true if some new text was added to the buffer, or false if the
1289buffer has reached the end of the input text.
1290
1291=cut
1292*/
1293
1294#define LEX_FAKE_EOF 0x80000000
112d1284 1295#define LEX_NO_TERM 0x40000000
f0e67a1d
Z
1296
1297bool
1298Perl_lex_next_chunk(pTHX_ U32 flags)
1299{
1300 SV *linestr;
1301 char *buf;
1302 STRLEN old_bufend_pos, new_bufend_pos;
1303 STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
1304 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
17cc9359 1305 bool got_some_for_debugger = 0;
f0e67a1d 1306 bool got_some;
112d1284 1307 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF|LEX_NO_TERM))
f0e67a1d 1308 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
f0e67a1d
Z
1309 linestr = PL_parser->linestr;
1310 buf = SvPVX(linestr);
1311 if (!(flags & LEX_KEEP_PREVIOUS) &&
1312 PL_parser->bufptr == PL_parser->bufend) {
1313 old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0;
1314 linestart_pos = 0;
1315 if (PL_parser->last_uni != PL_parser->bufend)
1316 PL_parser->last_uni = NULL;
1317 if (PL_parser->last_lop != PL_parser->bufend)
1318 PL_parser->last_lop = NULL;
1319 last_uni_pos = last_lop_pos = 0;
1320 *buf = 0;
1321 SvCUR(linestr) = 0;
1322 } else {
1323 old_bufend_pos = PL_parser->bufend - buf;
1324 bufptr_pos = PL_parser->bufptr - buf;
1325 oldbufptr_pos = PL_parser->oldbufptr - buf;
1326 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
1327 linestart_pos = PL_parser->linestart - buf;
1328 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
1329 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
1330 }
1331 if (flags & LEX_FAKE_EOF) {
1332 goto eof;
60d63348 1333 } else if (!PL_parser->rsfp && !PL_parser->filtered) {
f0e67a1d
Z
1334 got_some = 0;
1335 } else if (filter_gets(linestr, old_bufend_pos)) {
1336 got_some = 1;
17cc9359 1337 got_some_for_debugger = 1;
112d1284
FC
1338 } else if (flags & LEX_NO_TERM) {
1339 got_some = 0;
f0e67a1d 1340 } else {
580561a3
Z
1341 if (!SvPOK(linestr)) /* can get undefined by filter_gets */
1342 sv_setpvs(linestr, "");
f0e67a1d
Z
1343 eof:
1344 /* End of real input. Close filehandle (unless it was STDIN),
1345 * then add implicit termination.
1346 */
87606032 1347 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
f0e67a1d
Z
1348 PerlIO_clearerr(PL_parser->rsfp);
1349 else if (PL_parser->rsfp)
1350 (void)PerlIO_close(PL_parser->rsfp);
1351 PL_parser->rsfp = NULL;
60d63348 1352 PL_parser->in_pod = PL_parser->filtered = 0;
f0e67a1d
Z
1353#ifdef PERL_MAD
1354 if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n))
1355 PL_faketokens = 1;
1356#endif
1357 if (!PL_in_eval && PL_minus_p) {
1358 sv_catpvs(linestr,
1359 /*{*/";}continue{print or die qq(-p destination: $!\\n);}");
1360 PL_minus_n = PL_minus_p = 0;
1361 } else if (!PL_in_eval && PL_minus_n) {
1362 sv_catpvs(linestr, /*{*/";}");
1363 PL_minus_n = 0;
1364 } else
1365 sv_catpvs(linestr, ";");
1366 got_some = 1;
1367 }
1368 buf = SvPVX(linestr);
1369 new_bufend_pos = SvCUR(linestr);
1370 PL_parser->bufend = buf + new_bufend_pos;
1371 PL_parser->bufptr = buf + bufptr_pos;
1372 PL_parser->oldbufptr = buf + oldbufptr_pos;
1373 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
1374 PL_parser->linestart = buf + linestart_pos;
1375 if (PL_parser->last_uni)
1376 PL_parser->last_uni = buf + last_uni_pos;
1377 if (PL_parser->last_lop)
1378 PL_parser->last_lop = buf + last_lop_pos;
17cc9359 1379 if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) &&
f0e67a1d
Z
1380 PL_curstash != PL_debstash) {
1381 /* debugger active and we're not compiling the debugger code,
1382 * so store the line into the debugger's array of lines
1383 */
1384 update_debugger_info(NULL, buf+old_bufend_pos,
1385 new_bufend_pos-old_bufend_pos);
1386 }
1387 return got_some;
1388}
1389
1390/*
1391=for apidoc Amx|I32|lex_peek_unichar|U32 flags
1392
1393Looks ahead one (Unicode) character in the text currently being lexed.
1394Returns the codepoint (unsigned integer value) of the next character,
1395or -1 if lexing has reached the end of the input text. To consume the
1396peeked character, use L</lex_read_unichar>.
1397
1398If the next character is in (or extends into) the next chunk of input
1399text, the next chunk will be read in. Normally the current chunk will be
1400discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1401then the current chunk will not be discarded.
1402
1403If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1404is encountered, an exception is generated.
1405
1406=cut
1407*/
1408
1409I32
1410Perl_lex_peek_unichar(pTHX_ U32 flags)
1411{
749123ff 1412 dVAR;
f0e67a1d
Z
1413 char *s, *bufend;
1414 if (flags & ~(LEX_KEEP_PREVIOUS))
1415 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar");
1416 s = PL_parser->bufptr;
1417 bufend = PL_parser->bufend;
1418 if (UTF) {
1419 U8 head;
1420 I32 unichar;
1421 STRLEN len, retlen;
1422 if (s == bufend) {
1423 if (!lex_next_chunk(flags))
1424 return -1;
1425 s = PL_parser->bufptr;
1426 bufend = PL_parser->bufend;
1427 }
1428 head = (U8)*s;
54d004e8 1429 if (UTF8_IS_INVARIANT(head))
f0e67a1d 1430 return head;
54d004e8
KW
1431 if (UTF8_IS_START(head)) {
1432 len = UTF8SKIP(&head);
f0e67a1d
Z
1433 while ((STRLEN)(bufend-s) < len) {
1434 if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS))
1435 break;
1436 s = PL_parser->bufptr;
1437 bufend = PL_parser->bufend;
1438 }
1439 }
c80e42f3 1440 unichar = utf8n_to_uvchr((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY);
f0e67a1d
Z
1441 if (retlen == (STRLEN)-1) {
1442 /* malformed UTF-8 */
1443 ENTER;
1444 SAVESPTR(PL_warnhook);
1445 PL_warnhook = PERL_WARNHOOK_FATAL;
c80e42f3 1446 utf8n_to_uvchr((U8*)s, bufend-s, NULL, 0);
f0e67a1d
Z
1447 LEAVE;
1448 }
1449 return unichar;
1450 } else {
1451 if (s == bufend) {
1452 if (!lex_next_chunk(flags))
1453 return -1;
1454 s = PL_parser->bufptr;
1455 }
1456 return (U8)*s;
1457 }
1458}
1459
1460/*
1461=for apidoc Amx|I32|lex_read_unichar|U32 flags
1462
1463Reads the next (Unicode) character in the text currently being lexed.
1464Returns the codepoint (unsigned integer value) of the character read,
1465and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1
1466if lexing has reached the end of the input text. To non-destructively
1467examine the next character, use L</lex_peek_unichar> instead.
1468
1469If the next character is in (or extends into) the next chunk of input
1470text, the next chunk will be read in. Normally the current chunk will be
1471discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1472then the current chunk will not be discarded.
1473
1474If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1475is encountered, an exception is generated.
1476
1477=cut
1478*/
1479
1480I32
1481Perl_lex_read_unichar(pTHX_ U32 flags)
1482{
1483 I32 c;
1484 if (flags & ~(LEX_KEEP_PREVIOUS))
1485 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar");
1486 c = lex_peek_unichar(flags);
1487 if (c != -1) {
1488 if (c == '\n')
83944c01 1489 COPLINE_INC_WITH_HERELINES;
d9018cbe
EB
1490 if (UTF)
1491 PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr);
1492 else
1493 ++(PL_parser->bufptr);
f0e67a1d
Z
1494 }
1495 return c;
1496}
1497
1498/*
1499=for apidoc Amx|void|lex_read_space|U32 flags
1500
1501Reads optional spaces, in Perl style, in the text currently being
1502lexed. The spaces may include ordinary whitespace characters and
1503Perl-style comments. C<#line> directives are processed if encountered.
1504L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points
1505at a non-space character (or the end of the input text).
1506
1507If spaces extend into the next chunk of input text, the next chunk will
1508be read in. Normally the current chunk will be discarded at the same
1509time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current
1510chunk will not be discarded.
1511
1512=cut
1513*/
1514
21791330 1515#define LEX_NO_INCLINE 0x40000000
f0998909
Z
1516#define LEX_NO_NEXT_CHUNK 0x80000000
1517
f0e67a1d
Z
1518void
1519Perl_lex_read_space(pTHX_ U32 flags)
1520{
1521 char *s, *bufend;
21791330 1522 const bool can_incline = !(flags & LEX_NO_INCLINE);
f0e67a1d 1523 bool need_incline = 0;
21791330 1524 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK|LEX_NO_INCLINE))
f0e67a1d
Z
1525 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space");
1526#ifdef PERL_MAD
1527 if (PL_skipwhite) {
1528 sv_free(PL_skipwhite);
1529 PL_skipwhite = NULL;
1530 }
1531 if (PL_madskills)
1532 PL_skipwhite = newSVpvs("");
1533#endif /* PERL_MAD */
1534 s = PL_parser->bufptr;
1535 bufend = PL_parser->bufend;
1536 while (1) {
1537 char c = *s;
1538 if (c == '#') {
1539 do {
1540 c = *++s;
1541 } while (!(c == '\n' || (c == 0 && s == bufend)));
1542 } else if (c == '\n') {
1543 s++;
21791330
FC
1544 if (can_incline) {
1545 PL_parser->linestart = s;
1546 if (s == bufend)
1547 need_incline = 1;
1548 else
1549 incline(s);
1550 }
f0e67a1d
Z
1551 } else if (isSPACE(c)) {
1552 s++;
1553 } else if (c == 0 && s == bufend) {
1554 bool got_more;
1555#ifdef PERL_MAD
1556 if (PL_madskills)
1557 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1558#endif /* PERL_MAD */
f0998909
Z
1559 if (flags & LEX_NO_NEXT_CHUNK)
1560 break;
f0e67a1d 1561 PL_parser->bufptr = s;
21791330 1562 if (can_incline) COPLINE_INC_WITH_HERELINES;
f0e67a1d 1563 got_more = lex_next_chunk(flags);
21791330 1564 if (can_incline) CopLINE_dec(PL_curcop);
f0e67a1d
Z
1565 s = PL_parser->bufptr;
1566 bufend = PL_parser->bufend;
1567 if (!got_more)
1568 break;
21791330 1569 if (can_incline && need_incline && PL_parser->rsfp) {
f0e67a1d
Z
1570 incline(s);
1571 need_incline = 0;
1572 }
1573 } else {
1574 break;
1575 }
1576 }
1577#ifdef PERL_MAD
1578 if (PL_madskills)
1579 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1580#endif /* PERL_MAD */
1581 PL_parser->bufptr = s;
1582}
1583
1584/*
fe788d6b
PM
1585
1586=for apidoc EXMp|bool|validate_proto|SV *name|SV *proto|bool warn
1587
1588This function performs syntax checking on a prototype, C<proto>.
1589If C<warn> is true, any illegal characters or mismatched brackets
1590will trigger illegalproto warnings, declaring that they were
1591detected in the prototype for C<name>.
1592
1593The return value is C<true> if this is a valid prototype, and
1594C<false> if it is not, regardless of whether C<warn> was C<true> or
1595C<false>.
1596
1597Note that C<NULL> is a valid C<proto> and will always return C<true>.
1598
1599=cut
1600
1601 */
1602
1603bool
1604Perl_validate_proto(pTHX_ SV *name, SV *proto, bool warn)
1605{
1606 STRLEN len, origlen;
1607 char *p = proto ? SvPV(proto, len) : NULL;
1608 bool bad_proto = FALSE;
1609 bool in_brackets = FALSE;
1610 bool after_slash = FALSE;
1611 char greedy_proto = ' ';
1612 bool proto_after_greedy_proto = FALSE;
1613 bool must_be_last = FALSE;
1614 bool underscore = FALSE;
f791a21a 1615 bool bad_proto_after_underscore = FALSE;
fe788d6b
PM
1616
1617 PERL_ARGS_ASSERT_VALIDATE_PROTO;
1618
1619 if (!proto)
1620 return TRUE;
1621
1622 origlen = len;
1623 for (; len--; p++) {
1624 if (!isSPACE(*p)) {
1625 if (must_be_last)
1626 proto_after_greedy_proto = TRUE;
f791a21a
PM
1627 if (underscore) {
1628 if (!strchr(";@%", *p))
1629 bad_proto_after_underscore = TRUE;
1630 underscore = FALSE;
1631 }
fe788d6b
PM
1632 if (!strchr("$@%*;[]&\\_+", *p) || *p == '\0') {
1633 bad_proto = TRUE;
1634 }
1635 else {
fe788d6b
PM
1636 if (*p == '[')
1637 in_brackets = TRUE;
1638 else if (*p == ']')
1639 in_brackets = FALSE;
1640 else if ((*p == '@' || *p == '%') &&
1641 !after_slash &&
1642 !in_brackets ) {
1643 must_be_last = TRUE;
1644 greedy_proto = *p;
1645 }
1646 else if (*p == '_')
f791a21a 1647 underscore = TRUE;
fe788d6b
PM
1648 }
1649 if (*p == '\\')
1650 after_slash = TRUE;
1651 else
1652 after_slash = FALSE;
1653 }
1654 }
1655
1656 if (warn) {
b54d603d 1657 SV *tmpsv = newSVpvs_flags("", SVs_TEMP);
fe788d6b 1658 p -= origlen;
b54d603d
PM
1659 p = SvUTF8(proto)
1660 ? sv_uni_display(tmpsv, newSVpvn_flags(p, origlen, SVs_TEMP | SVf_UTF8),
1661 origlen, UNI_DISPLAY_ISPRINT)
1662 : pv_pretty(tmpsv, p, origlen, 60, NULL, NULL, PERL_PV_ESCAPE_NONASCII);
1663
fe788d6b
PM
1664 if (proto_after_greedy_proto)
1665 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1666 "Prototype after '%c' for %"SVf" : %s",
1667 greedy_proto, SVfARG(name), p);
50278ed0
PM
1668 if (in_brackets)
1669 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1670 "Missing ']' in prototype for %"SVf" : %s",
1671 SVfARG(name), p);
b54d603d 1672 if (bad_proto)
fe788d6b 1673 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
f791a21a
PM
1674 "Illegal character in prototype for %"SVf" : %s",
1675 SVfARG(name), p);
1676 if (bad_proto_after_underscore)
1677 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1678 "Illegal character after '_' in prototype for %"SVf" : %s",
1679 SVfARG(name), p);
fe788d6b
PM
1680 }
1681
1682 return (! (proto_after_greedy_proto || bad_proto) );
1683}
1684
1685/*
ffb4593c
NT
1686 * S_incline
1687 * This subroutine has nothing to do with tilting, whether at windmills
1688 * or pinball tables. Its name is short for "increment line". It
57843af0 1689 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 1690 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
1691 * # line 500 "foo.pm"
1692 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
1693 */
1694
76e3520e 1695STATIC void
d9095cec 1696S_incline(pTHX_ const char *s)
463ee0b2 1697{
97aff369 1698 dVAR;
d9095cec
NC
1699 const char *t;
1700 const char *n;
1701 const char *e;
8818d409 1702 line_t line_num;
463ee0b2 1703
7918f24d
NC
1704 PERL_ARGS_ASSERT_INCLINE;
1705
83944c01 1706 COPLINE_INC_WITH_HERELINES;
451f421f
FC
1707 if (!PL_rsfp && !PL_parser->filtered && PL_lex_state == LEX_NORMAL
1708 && s+1 == PL_bufend && *s == ';') {
1709 /* fake newline in string eval */
1710 CopLINE_dec(PL_curcop);
1711 return;
1712 }
463ee0b2
LW
1713 if (*s++ != '#')
1714 return;
d4c19fe8
AL
1715 while (SPACE_OR_TAB(*s))
1716 s++;
73659bf1
GS
1717 if (strnEQ(s, "line", 4))
1718 s += 4;
1719 else
1720 return;
084592ab 1721 if (SPACE_OR_TAB(*s))
73659bf1 1722 s++;
4e553d73 1723 else
73659bf1 1724 return;
d4c19fe8
AL
1725 while (SPACE_OR_TAB(*s))
1726 s++;
463ee0b2
LW
1727 if (!isDIGIT(*s))
1728 return;
d4c19fe8 1729
463ee0b2
LW
1730 n = s;
1731 while (isDIGIT(*s))
1732 s++;
07714eb4 1733 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
26b6dc3f 1734 return;
bf4acbe4 1735 while (SPACE_OR_TAB(*s))
463ee0b2 1736 s++;
73659bf1 1737 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 1738 s++;
73659bf1
GS
1739 e = t + 1;
1740 }
463ee0b2 1741 else {
c35e046a
AL
1742 t = s;
1743 while (!isSPACE(*t))
1744 t++;
73659bf1 1745 e = t;
463ee0b2 1746 }
bf4acbe4 1747 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
1748 e++;
1749 if (*e != '\n' && *e != '\0')
1750 return; /* false alarm */
1751
8818d409
FC
1752 line_num = atoi(n)-1;
1753
f4dd75d9 1754 if (t - s > 0) {
d9095cec 1755 const STRLEN len = t - s;
3df32bda 1756
d36ee5be 1757 if (!PL_rsfp && !PL_parser->filtered) {
e66cf94c
RGS
1758 /* must copy *{"::_<(eval N)[oldfilename:L]"}
1759 * to *{"::_<newfilename"} */
44867030
NC
1760 /* However, the long form of evals is only turned on by the
1761 debugger - usually they're "(eval %lu)" */
d36ee5be
FC
1762 GV * const cfgv = CopFILEGV(PL_curcop);
1763 if (cfgv) {
38bd7ad8
FC
1764 char smallbuf[128];
1765 STRLEN tmplen2 = len;
44867030 1766 char *tmpbuf2;
449dd039 1767 GV *gv2;
44867030
NC
1768
1769 if (tmplen2 + 2 <= sizeof smallbuf)
1770 tmpbuf2 = smallbuf;
1771 else
1772 Newx(tmpbuf2, tmplen2 + 2, char);
1773
38bd7ad8
FC
1774 tmpbuf2[0] = '_';
1775 tmpbuf2[1] = '<';
44867030
NC
1776
1777 memcpy(tmpbuf2 + 2, s, tmplen2);
1778 tmplen2 += 2;
1779
8a5ee598 1780 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
e5527e4b 1781 if (!isGV(gv2)) {
8a5ee598 1782 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
e5527e4b
RGS
1783 /* adjust ${"::_<newfilename"} to store the new file name */
1784 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
8818d409
FC
1785 /* The line number may differ. If that is the case,
1786 alias the saved lines that are in the array.
1787 Otherwise alias the whole array. */
1788 if (CopLINE(PL_curcop) == line_num) {
38bd7ad8
FC
1789 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(cfgv)));
1790 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(cfgv)));
8818d409 1791 }
38bd7ad8
FC
1792 else if (GvAV(cfgv)) {
1793 AV * const av = GvAV(cfgv);
8818d409
FC
1794 const I32 start = CopLINE(PL_curcop)+1;
1795 I32 items = AvFILLp(av) - start;
1796 if (items > 0) {
1797 AV * const av2 = GvAVn(gv2);
1798 SV **svp = AvARRAY(av) + start;
1799 I32 l = (I32)line_num+1;
1800 while (items--)
1801 av_store(av2, l++, SvREFCNT_inc(*svp++));
1802 }
1803 }
e5527e4b 1804 }
44867030
NC
1805
1806 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
d36ee5be 1807 }
e66cf94c 1808 }
05ec9bb3 1809 CopFILE_free(PL_curcop);
449dd039 1810 CopFILE_setn(PL_curcop, s, len);
f4dd75d9 1811 }
8818d409 1812 CopLINE_set(PL_curcop, line_num);
463ee0b2
LW
1813}
1814
21791330
FC
1815#define skipspace(s) skipspace_flags(s, 0)
1816
29595ff2 1817#ifdef PERL_MAD
cd81e915 1818/* skip space before PL_thistoken */
29595ff2
NC
1819
1820STATIC char *
5aaab254 1821S_skipspace0(pTHX_ char *s)
29595ff2 1822{
7918f24d
NC
1823 PERL_ARGS_ASSERT_SKIPSPACE0;
1824
29595ff2
NC
1825 s = skipspace(s);
1826 if (!PL_madskills)
1827 return s;
cd81e915
NC
1828 if (PL_skipwhite) {
1829 if (!PL_thiswhite)
6b29d1f5 1830 PL_thiswhite = newSVpvs("");
cd81e915
NC
1831 sv_catsv(PL_thiswhite, PL_skipwhite);
1832 sv_free(PL_skipwhite);
1833 PL_skipwhite = 0;
1834 }
1835 PL_realtokenstart = s - SvPVX(PL_linestr);
29595ff2
NC
1836 return s;
1837}
1838
cd81e915 1839/* skip space after PL_thistoken */
29595ff2
NC
1840
1841STATIC char *
5aaab254 1842S_skipspace1(pTHX_ char *s)
29595ff2 1843{
d4c19fe8 1844 const char *start = s;
29595ff2
NC
1845 I32 startoff = start - SvPVX(PL_linestr);
1846
7918f24d
NC
1847 PERL_ARGS_ASSERT_SKIPSPACE1;
1848
29595ff2
NC
1849 s = skipspace(s);
1850 if (!PL_madskills)
1851 return s;
1852 start = SvPVX(PL_linestr) + startoff;
cd81e915 1853 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1854 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1855 PL_thistoken = newSVpvn(tstart, start - tstart);
1856 }
1857 PL_realtokenstart = -1;
1858 if (PL_skipwhite) {
1859 if (!PL_nextwhite)
6b29d1f5 1860 PL_nextwhite = newSVpvs("");
cd81e915
NC
1861 sv_catsv(PL_nextwhite, PL_skipwhite);
1862 sv_free(PL_skipwhite);
1863 PL_skipwhite = 0;
29595ff2
NC
1864 }
1865 return s;
1866}
1867
1868STATIC char *
5aaab254 1869S_skipspace2(pTHX_ char *s, SV **svp)
29595ff2 1870{
c35e046a 1871 char *start;
c35e046a
AL
1872 const I32 startoff = s - SvPVX(PL_linestr);
1873
7918f24d
NC
1874 PERL_ARGS_ASSERT_SKIPSPACE2;
1875
29595ff2 1876 s = skipspace(s);
29595ff2
NC
1877 if (!PL_madskills || !svp)
1878 return s;
1879 start = SvPVX(PL_linestr) + startoff;
cd81e915 1880 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1881 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1882 PL_thistoken = newSVpvn(tstart, start - tstart);
1883 PL_realtokenstart = -1;
29595ff2 1884 }
cd81e915 1885 if (PL_skipwhite) {
29595ff2 1886 if (!*svp)
6b29d1f5 1887 *svp = newSVpvs("");
cd81e915
NC
1888 sv_setsv(*svp, PL_skipwhite);
1889 sv_free(PL_skipwhite);
1890 PL_skipwhite = 0;
29595ff2
NC
1891 }
1892
1893 return s;
1894}
1895#endif
1896
80a702cd 1897STATIC void
15f169a1 1898S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
80a702cd
RGS
1899{
1900 AV *av = CopFILEAVx(PL_curcop);
1901 if (av) {
b9f83d2f 1902 SV * const sv = newSV_type(SVt_PVMG);
5fa550fb 1903 if (orig_sv)
4e917a04 1904 sv_setsv_flags(sv, orig_sv, 0); /* no cow */
5fa550fb
NC
1905 else
1906 sv_setpvn(sv, buf, len);
80a702cd
RGS
1907 (void)SvIOK_on(sv);
1908 SvIV_set(sv, 0);
c70927a6 1909 av_store(av, CopLINE(PL_curcop), sv);
80a702cd
RGS
1910 }
1911}
1912
ffb4593c
NT
1913/*
1914 * S_skipspace
1915 * Called to gobble the appropriate amount and type of whitespace.
1916 * Skips comments as well.
1917 */
1918
76e3520e 1919STATIC char *
21791330 1920S_skipspace_flags(pTHX_ char *s, U32 flags)
a687059c 1921{
5db06880 1922#ifdef PERL_MAD
f0e67a1d
Z
1923 char *start = s;
1924#endif /* PERL_MAD */
21791330 1925 PERL_ARGS_ASSERT_SKIPSPACE_FLAGS;
f0e67a1d 1926#ifdef PERL_MAD
cd81e915
NC
1927 if (PL_skipwhite) {
1928 sv_free(PL_skipwhite);
f0e67a1d 1929 PL_skipwhite = NULL;
5db06880 1930 }
f0e67a1d 1931#endif /* PERL_MAD */
3280af22 1932 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 1933 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2 1934 s++;
f0e67a1d
Z
1935 } else {
1936 STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1937 PL_bufptr = s;
21791330 1938 lex_read_space(flags | LEX_KEEP_PREVIOUS |
f0998909
Z
1939 (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1940 LEX_NO_NEXT_CHUNK : 0));
3280af22 1941 s = PL_bufptr;
f0e67a1d
Z
1942 PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1943 if (PL_linestart > PL_bufptr)
1944 PL_bufptr = PL_linestart;
1945 return s;
463ee0b2 1946 }
5db06880 1947#ifdef PERL_MAD
f0e67a1d
Z
1948 if (PL_madskills)
1949 PL_skipwhite = newSVpvn(start, s-start);
1950#endif /* PERL_MAD */
5db06880 1951 return s;
a687059c 1952}
378cc40b 1953
ffb4593c
NT
1954/*
1955 * S_check_uni
1956 * Check the unary operators to ensure there's no ambiguity in how they're
1957 * used. An ambiguous piece of code would be:
1958 * rand + 5
1959 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1960 * the +5 is its argument.
1961 */
1962
76e3520e 1963STATIC void
cea2e8a9 1964S_check_uni(pTHX)
ba106d47 1965{
97aff369 1966 dVAR;
d4c19fe8
AL
1967 const char *s;
1968 const char *t;
2f3197b3 1969
3280af22 1970 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 1971 return;
3280af22
NIS
1972 while (isSPACE(*PL_last_uni))
1973 PL_last_uni++;
c35e046a 1974 s = PL_last_uni;
8a2bca12 1975 while (isWORDCHAR_lazy_if(s,UTF) || *s == '-')
c35e046a 1976 s++;
3280af22 1977 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 1978 return;
6136c704 1979
9b387841
NC
1980 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
1981 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1982 (int)(s - PL_last_uni), PL_last_uni);
2f3197b3
LW
1983}
1984
ffb4593c
NT
1985/*
1986 * LOP : macro to build a list operator. Its behaviour has been replaced
1987 * with a subroutine, S_lop() for which LOP is just another name.
1988 */
1989
a0d0e21e
LW
1990#define LOP(f,x) return lop(f,x,s)
1991
ffb4593c
NT
1992/*
1993 * S_lop
1994 * Build a list operator (or something that might be one). The rules:
1995 * - if we have a next token, then it's a list operator [why?]
1996 * - if the next thing is an opening paren, then it's a function
1997 * - else it's a list operator
1998 */
1999
76e3520e 2000STATIC I32
a0be28da 2001S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 2002{
97aff369 2003 dVAR;
7918f24d
NC
2004
2005 PERL_ARGS_ASSERT_LOP;
2006
6154021b 2007 pl_yylval.ival = f;
35c8bce7 2008 CLINE;
3280af22
NIS
2009 PL_expect = x;
2010 PL_bufptr = s;
2011 PL_last_lop = PL_oldbufptr;
eb160463 2012 PL_last_lop_op = (OPCODE)f;
5db06880
NC
2013#ifdef PERL_MAD
2014 if (PL_lasttoke)
78cdf107 2015 goto lstop;
5db06880 2016#else
3280af22 2017 if (PL_nexttoke)
78cdf107 2018 goto lstop;
5db06880 2019#endif
79072805 2020 if (*s == '(')
bbf60fe6 2021 return REPORT(FUNC);
29595ff2 2022 s = PEEKSPACE(s);
79072805 2023 if (*s == '(')
bbf60fe6 2024 return REPORT(FUNC);
78cdf107
Z
2025 else {
2026 lstop:
2027 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
2028 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
bbf60fe6 2029 return REPORT(LSTOP);
78cdf107 2030 }
79072805
LW
2031}
2032
5db06880
NC
2033#ifdef PERL_MAD
2034 /*
2035 * S_start_force
2036 * Sets up for an eventual force_next(). start_force(0) basically does
2037 * an unshift, while start_force(-1) does a push. yylex removes items
2038 * on the "pop" end.
2039 */
2040
2041STATIC void
2042S_start_force(pTHX_ int where)
2043{
2044 int i;
2045
cd81e915 2046 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
5db06880 2047 where = PL_lasttoke;
cd81e915
NC
2048 assert(PL_curforce < 0 || PL_curforce == where);
2049 if (PL_curforce != where) {
5db06880
NC
2050 for (i = PL_lasttoke; i > where; --i) {
2051 PL_nexttoke[i] = PL_nexttoke[i-1];
2052 }
2053 PL_lasttoke++;
2054 }
cd81e915 2055 if (PL_curforce < 0) /* in case of duplicate start_force() */
5db06880 2056 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
cd81e915
NC
2057 PL_curforce = where;
2058 if (PL_nextwhite) {
5db06880 2059 if (PL_madskills)
6b29d1f5 2060 curmad('^', newSVpvs(""));
cd81e915 2061 CURMAD('_', PL_nextwhite);
5db06880
NC
2062 }
2063}
2064
2065STATIC void
2066S_curmad(pTHX_ char slot, SV *sv)
2067{
2068 MADPROP **where;
2069
2070 if (!sv)
2071 return;
cd81e915
NC
2072 if (PL_curforce < 0)
2073 where = &PL_thismad;
5db06880 2074 else
cd81e915 2075 where = &PL_nexttoke[PL_curforce].next_mad;
5db06880 2076
cd81e915 2077 if (PL_faketokens)
76f68e9b 2078 sv_setpvs(sv, "");
5db06880
NC
2079 else {
2080 if (!IN_BYTES) {
2081 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2082 SvUTF8_on(sv);
2083 else if (PL_encoding) {
2084 sv_recode_to_utf8(sv, PL_encoding);
2085 }
2086 }
2087 }
2088
2089 /* keep a slot open for the head of the list? */
2090 if (slot != '_' && *where && (*where)->mad_key == '^') {
2091 (*where)->mad_key = slot;
daba3364 2092 sv_free(MUTABLE_SV(((*where)->mad_val)));
5db06880
NC
2093 (*where)->mad_val = (void*)sv;
2094 }
2095 else
2096 addmad(newMADsv(slot, sv), where, 0);
2097}
2098#else
b3f24c00
MHM
2099# define start_force(where) NOOP
2100# define curmad(slot, sv) NOOP
5db06880
NC
2101#endif
2102
ffb4593c
NT
2103/*
2104 * S_force_next
9cbb5ea2 2105 * When the lexer realizes it knows the next token (for instance,
ffb4593c 2106 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2 2107 * to know what token to return the next time the lexer is called. Caller
5db06880
NC
2108 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
2109 * and possibly PL_expect to ensure the lexer handles the token correctly.
ffb4593c
NT
2110 */
2111
4e553d73 2112STATIC void
cea2e8a9 2113S_force_next(pTHX_ I32 type)
79072805 2114{
97aff369 2115 dVAR;
704d4215
GG
2116#ifdef DEBUGGING
2117 if (DEBUG_T_TEST) {
2118 PerlIO_printf(Perl_debug_log, "### forced token:\n");
f05d7009 2119 tokereport(type, &NEXTVAL_NEXTTOKE);
704d4215
GG
2120 }
2121#endif
5db06880 2122#ifdef PERL_MAD
cd81e915 2123 if (PL_curforce < 0)
5db06880 2124 start_force(PL_lasttoke);
cd81e915 2125 PL_nexttoke[PL_curforce].next_type = type;
5db06880
NC
2126 if (PL_lex_state != LEX_KNOWNEXT)
2127 PL_lex_defer = PL_lex_state;
2128 PL_lex_state = LEX_KNOWNEXT;
2129 PL_lex_expect = PL_expect;
cd81e915 2130 PL_curforce = -1;
5db06880 2131#else
3280af22
NIS
2132 PL_nexttype[PL_nexttoke] = type;
2133 PL_nexttoke++;
2134 if (PL_lex_state != LEX_KNOWNEXT) {
2135 PL_lex_defer = PL_lex_state;
2136 PL_lex_expect = PL_expect;
2137 PL_lex_state = LEX_KNOWNEXT;
79072805 2138 }
5db06880 2139#endif
79072805
LW
2140}
2141
28ac2b49
Z
2142void
2143Perl_yyunlex(pTHX)
2144{
a7aaec61
Z
2145 int yyc = PL_parser->yychar;
2146 if (yyc != YYEMPTY) {
2147 if (yyc) {
2148 start_force(-1);
2149 NEXTVAL_NEXTTOKE = PL_parser->yylval;
2150 if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) {
78cdf107 2151 PL_lex_allbrackets--;
a7aaec61 2152 PL_lex_brackets--;
78cdf107
Z
2153 yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16);
2154 } else if (yyc == '('/*)*/) {
2155 PL_lex_allbrackets--;
2156 yyc |= (2<<24);
a7aaec61
Z
2157 }
2158 force_next(yyc);
2159 }
28ac2b49
Z
2160 PL_parser->yychar = YYEMPTY;
2161 }
2162}
2163
d0a148a6 2164STATIC SV *
15f169a1 2165S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
d0a148a6 2166{
97aff369 2167 dVAR;
740cce10 2168 SV * const sv = newSVpvn_utf8(start, len,
eaf7a4d2
CS
2169 !IN_BYTES
2170 && UTF
2171 && !is_ascii_string((const U8*)start, len)
740cce10 2172 && is_utf8_string((const U8*)start, len));
d0a148a6
NC
2173 return sv;
2174}
2175
ffb4593c
NT
2176/*
2177 * S_force_word
2178 * When the lexer knows the next thing is a word (for instance, it has
2179 * just seen -> and it knows that the next char is a word char, then
02b34bbe
DM
2180 * it calls S_force_word to stick the next word into the PL_nexttoke/val
2181 * lookahead.
ffb4593c
NT
2182 *
2183 * Arguments:
b1b65b59 2184 * char *start : buffer position (must be within PL_linestr)
02b34bbe 2185 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
ffb4593c
NT
2186 * int check_keyword : if true, Perl checks to make sure the word isn't
2187 * a keyword (do this if the word is a label, e.g. goto FOO)
2188 * int allow_pack : if true, : characters will also be allowed (require,
2189 * use, etc. do this)
9cbb5ea2 2190 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
2191 */
2192
76e3520e 2193STATIC char *
345b3785 2194S_force_word(pTHX_ char *start, int token, int check_keyword, int allow_pack)
79072805 2195{
97aff369 2196 dVAR;
eb578fdb 2197 char *s;
463ee0b2 2198 STRLEN len;
4e553d73 2199
7918f24d
NC
2200 PERL_ARGS_ASSERT_FORCE_WORD;
2201
29595ff2 2202 start = SKIPSPACE1(start);
463ee0b2 2203 s = start;
7e2040f0 2204 if (isIDFIRST_lazy_if(s,UTF) ||
345b3785 2205 (allow_pack && *s == ':') )
a0d0e21e 2206 {
3280af22 2207 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
01b5ef50
FC
2208 if (check_keyword) {
2209 char *s2 = PL_tokenbuf;
2210 if (allow_pack && len > 6 && strnEQ(s2, "CORE::", 6))
2211 s2 += 6, len -= 6;
2212 if (keyword(s2, len, 0))
463ee0b2 2213 return start;
01b5ef50 2214 }
cd81e915 2215 start_force(PL_curforce);
5db06880
NC
2216 if (PL_madskills)
2217 curmad('X', newSVpvn(start,s-start));
463ee0b2 2218 if (token == METHOD) {
29595ff2 2219 s = SKIPSPACE1(s);
463ee0b2 2220 if (*s == '(')
3280af22 2221 PL_expect = XTERM;
463ee0b2 2222 else {
3280af22 2223 PL_expect = XOPERATOR;
463ee0b2 2224 }
79072805 2225 }
e74e6b3d 2226 if (PL_madskills)
63575281 2227 curmad('g', newSVpvs( "forced" ));
9ded7720 2228 NEXTVAL_NEXTTOKE.opval
d0a148a6
NC
2229 = (OP*)newSVOP(OP_CONST,0,
2230 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
9ded7720 2231 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
79072805
LW
2232 force_next(token);
2233 }
2234 return s;
2235}
2236
ffb4593c
NT
2237/*
2238 * S_force_ident
9cbb5ea2 2239 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
2240 * text only contains the "foo" portion. The first argument is a pointer
2241 * to the "foo", and the second argument is the type symbol to prefix.
2242 * Forces the next token to be a "WORD".
9cbb5ea2 2243 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
2244 */
2245
76e3520e 2246STATIC void
5aaab254 2247S_force_ident(pTHX_ const char *s, int kind)
79072805 2248{
97aff369 2249 dVAR;
7918f24d
NC
2250
2251 PERL_ARGS_ASSERT_FORCE_IDENT;
2252
c9b48522
DD
2253 if (s[0]) {
2254 const STRLEN len = s[1] ? strlen(s) : 1; /* s = "\"" see yylex */
728847b1
BF
2255 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn_flags(s, len,
2256 UTF ? SVf_UTF8 : 0));
cd81e915 2257 start_force(PL_curforce);
9ded7720 2258 NEXTVAL_NEXTTOKE.opval = o;
79072805 2259 force_next(WORD);
748a9306 2260 if (kind) {
11343788 2261 o->op_private = OPpCONST_ENTERED;
55497cff
PP
2262 /* XXX see note in pp_entereval() for why we forgo typo
2263 warnings if the symbol must be introduced in an eval.
2264 GSAR 96-10-12 */
90e5519e 2265 gv_fetchpvn_flags(s, len,
728847b1
BF
2266 (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2267 : GV_ADD) | ( UTF ? SVf_UTF8 : 0 ),
90e5519e
NC
2268 kind == '$' ? SVt_PV :
2269 kind == '@' ? SVt_PVAV :
2270 kind == '%' ? SVt_PVHV :
a0d0e21e 2271 SVt_PVGV
90e5519e 2272 );
748a9306 2273 }
79072805
LW
2274 }
2275}
2276
3f33d153
FC
2277static void
2278S_force_ident_maybe_lex(pTHX_ char pit)
2279{
2280 start_force(PL_curforce);
2281 NEXTVAL_NEXTTOKE.ival = pit;
2282 force_next('p');
2283}
2284
1571675a
GS
2285NV
2286Perl_str_to_version(pTHX_ SV *sv)
2287{
2288 NV retval = 0.0;
2289 NV nshift = 1.0;
2290 STRLEN len;
cfd0369c 2291 const char *start = SvPV_const(sv,len);
9d4ba2ae 2292 const char * const end = start + len;
504618e9 2293 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
7918f24d
NC
2294
2295 PERL_ARGS_ASSERT_STR_TO_VERSION;
2296
1571675a 2297 while (start < end) {
ba210ebe 2298 STRLEN skip;
1571675a
GS
2299 UV n;
2300 if (utf)
9041c2e3 2301 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
2302 else {
2303 n = *(U8*)start;
2304 skip = 1;
2305 }
2306 retval += ((NV)n)/nshift;
2307 start += skip;
2308 nshift *= 1000;
2309 }
2310 return retval;
2311}
2312
4e553d73 2313/*
ffb4593c
NT
2314 * S_force_version
2315 * Forces the next token to be a version number.
e759cc13
RGS
2316 * If the next token appears to be an invalid version number, (e.g. "v2b"),
2317 * and if "guessing" is TRUE, then no new token is created (and the caller
2318 * must use an alternative parsing method).
ffb4593c
NT
2319 */
2320
76e3520e 2321STATIC char *
e759cc13 2322S_force_version(pTHX_ char *s, int guessing)
89bfa8cd 2323{
97aff369 2324 dVAR;
5f66b61c 2325 OP *version = NULL;
44dcb63b 2326 char *d;
5db06880
NC
2327#ifdef PERL_MAD
2328 I32 startoff = s - SvPVX(PL_linestr);
2329#endif
89bfa8cd 2330
7918f24d
NC
2331 PERL_ARGS_ASSERT_FORCE_VERSION;
2332
29595ff2 2333 s = SKIPSPACE1(s);
89bfa8cd 2334
44dcb63b 2335 d = s;
dd629d5b 2336 if (*d == 'v')
44dcb63b 2337 d++;
44dcb63b 2338 if (isDIGIT(*d)) {
e759cc13
RGS
2339 while (isDIGIT(*d) || *d == '_' || *d == '.')
2340 d++;
5db06880
NC
2341#ifdef PERL_MAD
2342 if (PL_madskills) {
cd81e915 2343 start_force(PL_curforce);
5db06880
NC
2344 curmad('X', newSVpvn(s,d-s));
2345 }
2346#endif
4e4da3ac 2347 if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) {
dd629d5b 2348 SV *ver;
8d08d9ba 2349#ifdef USE_LOCALE_NUMERIC
909d3787
KW
2350 char *loc = savepv(setlocale(LC_NUMERIC, NULL));
2351 setlocale(LC_NUMERIC, "C");
8d08d9ba 2352#endif
6154021b 2353 s = scan_num(s, &pl_yylval);
8d08d9ba
DG
2354#ifdef USE_LOCALE_NUMERIC
2355 setlocale(LC_NUMERIC, loc);
909d3787 2356 Safefree(loc);
8d08d9ba 2357#endif
6154021b 2358 version = pl_yylval.opval;
dd629d5b
GS
2359 ver = cSVOPx(version)->op_sv;
2360 if (SvPOK(ver) && !SvNIOK(ver)) {
862a34c6 2361 SvUPGRADE(ver, SVt_PVNV);
9d6ce603 2362 SvNV_set(ver, str_to_version(ver));
1571675a 2363 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 2364 }
89bfa8cd 2365 }
5db06880
NC
2366 else if (guessing) {
2367#ifdef PERL_MAD
2368 if (PL_madskills) {
cd81e915
NC
2369 sv_free(PL_nextwhite); /* let next token collect whitespace */
2370 PL_nextwhite = 0;
5db06880
NC
2371 s = SvPVX(PL_linestr) + startoff;
2372 }
2373#endif
e759cc13 2374 return s;
5db06880 2375 }
89bfa8cd
PP
2376 }
2377
5db06880
NC
2378#ifdef PERL_MAD
2379 if (PL_madskills && !version) {
cd81e915
NC
2380 sv_free(PL_nextwhite); /* let next token collect whitespace */
2381 PL_nextwhite = 0;
5db06880
NC
2382 s = SvPVX(PL_linestr) + startoff;
2383 }
2384#endif
89bfa8cd 2385 /* NOTE: The parser sees the package name and the VERSION swapped */
cd81e915 2386 start_force(PL_curforce);
9ded7720 2387 NEXTVAL_NEXTTOKE.opval = version;
4e553d73 2388 force_next(WORD);
89bfa8cd 2389
e759cc13 2390 return s;
89bfa8cd
PP
2391}
2392
ffb4593c 2393/*
91152fc1
DG
2394 * S_force_strict_version
2395 * Forces the next token to be a version number using strict syntax rules.
2396 */
2397
2398STATIC char *
2399S_force_strict_version(pTHX_ char *s)
2400{
2401 dVAR;
2402 OP *version = NULL;
2403#ifdef PERL_MAD
2404 I32 startoff = s - SvPVX(PL_linestr);
2405#endif
2406 const char *errstr = NULL;
2407
2408 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2409
2410 while (isSPACE(*s)) /* leading whitespace */
2411 s++;
2412
2413 if (is_STRICT_VERSION(s,&errstr)) {
2414 SV *ver = newSV(0);
2415 s = (char *)scan_version(s, ver, 0);
2416 version = newSVOP(OP_CONST, 0, ver);
2417 }
4e4da3ac
Z
2418 else if ( (*s != ';' && *s != '{' && *s != '}' ) &&
2419 (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' )))
2420 {
91152fc1
DG
2421 PL_bufptr = s;
2422 if (errstr)
2423 yyerror(errstr); /* version required */
2424 return s;
2425 }
2426
2427#ifdef PERL_MAD
2428 if (PL_madskills && !version) {
2429 sv_free(PL_nextwhite); /* let next token collect whitespace */
2430 PL_nextwhite = 0;
2431 s = SvPVX(PL_linestr) + startoff;
2432 }
2433#endif
2434 /* NOTE: The parser sees the package name and the VERSION swapped */
2435 start_force(PL_curforce);
2436 NEXTVAL_NEXTTOKE.opval = version;
2437 force_next(WORD);
2438
2439 return s;
2440}
2441
2442/*
ffb4593c
NT
2443 * S_tokeq
2444 * Tokenize a quoted string passed in as an SV. It finds the next
2445 * chunk, up to end of string or a backslash. It may make a new
2446 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2447 * turns \\ into \.
2448 */
2449
76e3520e 2450STATIC SV *
cea2e8a9 2451S_tokeq(pTHX_ SV *sv)
79072805 2452{
97aff369 2453 dVAR;
eb578fdb
KW
2454 char *s;
2455 char *send;
2456 char *d;
b3ac6de7
IZ
2457 STRLEN len = 0;
2458 SV *pv = sv;
79072805 2459
7918f24d
NC
2460 PERL_ARGS_ASSERT_TOKEQ;
2461
79072805 2462 if (!SvLEN(sv))
b3ac6de7 2463 goto finish;
79072805 2464
a0d0e21e 2465 s = SvPV_force(sv, len);
21a311ee 2466 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 2467 goto finish;
463ee0b2 2468 send = s + len;
dcb21ed6
NC
2469 /* This is relying on the SV being "well formed" with a trailing '\0' */
2470 while (s < send && !(*s == '\\' && s[1] == '\\'))
79072805
LW
2471 s++;
2472 if (s == send)
b3ac6de7 2473 goto finish;
79072805 2474 d = s;
be4731d2 2475 if ( PL_hints & HINT_NEW_STRING ) {
59cd0e26 2476 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
be4731d2 2477 }
79072805
LW
2478 while (s < send) {
2479 if (*s == '\\') {
a0d0e21e 2480 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
2481 s++; /* all that, just for this */
2482 }
2483 *d++ = *s++;
2484 }
2485 *d = '\0';
95a20fc0 2486 SvCUR_set(sv, d - SvPVX_const(sv));
b3ac6de7 2487 finish:
3280af22 2488 if ( PL_hints & HINT_NEW_STRING )
eb0d8d16 2489 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
79072805
LW
2490 return sv;
2491}
2492
ffb4593c
NT
2493/*
2494 * Now come three functions related to double-quote context,
2495 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2496 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2497 * interact with PL_lex_state, and create fake ( ... ) argument lists
2498 * to handle functions and concatenation.
ecd24171
DM
2499 * For example,
2500 * "foo\lbar"
2501 * is tokenised as
2502 * stringify ( const[foo] concat lcfirst ( const[bar] ) )
ffb4593c
NT
2503 */
2504
2505/*
2506 * S_sublex_start
6154021b 2507 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
ffb4593c
NT
2508 *
2509 * Pattern matching will set PL_lex_op to the pattern-matching op to
6154021b 2510 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
ffb4593c
NT
2511 *
2512 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2513 *
2514 * Everything else becomes a FUNC.
2515 *
2516 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2517 * had an OP_CONST or OP_READLINE). This just sets us up for a
2518 * call to S_sublex_push().
2519 */
2520
76e3520e 2521STATIC I32
cea2e8a9 2522S_sublex_start(pTHX)
79072805 2523{
97aff369 2524 dVAR;
eb578fdb 2525 const I32 op_type = pl_yylval.ival;
79072805
LW
2526
2527 if (op_type == OP_NULL) {
6154021b 2528 pl_yylval.opval = PL_lex_op;
5f66b61c 2529 PL_lex_op = NULL;
79072805
LW
2530 return THING;
2531 }
2532 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 2533 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
2534
2535 if (SvTYPE(sv) == SVt_PVIV) {
2536 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2537 STRLEN len;
96a5add6 2538 const char * const p = SvPV_const(sv, len);
740cce10 2539 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
b3ac6de7
IZ
2540 SvREFCNT_dec(sv);
2541 sv = nsv;
4e553d73 2542 }
6154021b 2543 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
a0714e2c 2544 PL_lex_stuff = NULL;
6f33ba73
RGS
2545 /* Allow <FH> // "foo" */
2546 if (op_type == OP_READLINE)
2547 PL_expect = XTERMORDORDOR;
79072805
LW
2548 return THING;
2549 }
e3f73d4e 2550 else if (op_type == OP_BACKTICK && PL_lex_op) {
06df4f41 2551 /* readpipe() was overridden */
e3f73d4e 2552 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
6154021b 2553 pl_yylval.opval = PL_lex_op;
9b201d7d 2554 PL_lex_op = NULL;
e3f73d4e
RGS
2555 PL_lex_stuff = NULL;
2556 return THING;
2557 }
79072805 2558
3280af22 2559 PL_sublex_info.super_state = PL_lex_state;
eac04b2e 2560 PL_sublex_info.sub_inwhat = (U16)op_type;
3280af22
NIS
2561 PL_sublex_info.sub_op = PL_lex_op;
2562 PL_lex_state = LEX_INTERPPUSH;
55497cff 2563
3280af22
NIS
2564 PL_expect = XTERM;
2565 if (PL_lex_op) {
6154021b 2566 pl_yylval.opval = PL_lex_op;
5f66b61c 2567 PL_lex_op = NULL;
55497cff
PP
2568 return PMFUNC;
2569 }
2570 else
2571 return FUNC;
2572}
2573
ffb4593c
NT
2574/*
2575 * S_sublex_push
2576 * Create a new scope to save the lexing state. The scope will be
2577 * ended in S_sublex_done. Returns a '(', starting the function arguments
2578 * to the uc, lc, etc. found before.
2579 * Sets PL_lex_state to LEX_INTERPCONCAT.
2580 */
2581
76e3520e 2582STATIC I32
cea2e8a9 2583S_sublex_push(pTHX)
55497cff 2584{
27da23d5 2585 dVAR;
78a635de 2586 LEXSHARED *shared;
f46d017c 2587 ENTER;
55497cff 2588
3280af22 2589 PL_lex_state = PL_sublex_info.super_state;
651b5b28 2590 SAVEBOOL(PL_lex_dojoin);
3280af22 2591 SAVEI32(PL_lex_brackets);
78cdf107 2592 SAVEI32(PL_lex_allbrackets);
b27dce25 2593 SAVEI32(PL_lex_formbrack);
78cdf107 2594 SAVEI8(PL_lex_fakeeof);
3280af22
NIS
2595 SAVEI32(PL_lex_casemods);
2596 SAVEI32(PL_lex_starts);
651b5b28 2597 SAVEI8(PL_lex_state);
7cc34111 2598 SAVESPTR(PL_lex_repl);
7766f137 2599 SAVEVPTR(PL_lex_inpat);
98246f1e 2600 SAVEI16(PL_lex_inwhat);
57843af0 2601 SAVECOPLINE(PL_curcop);
3280af22 2602 SAVEPPTR(PL_bufptr);
8452ff4b 2603 SAVEPPTR(PL_bufend);
3280af22
NIS
2604 SAVEPPTR(PL_oldbufptr);
2605 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
2606 SAVEPPTR(PL_last_lop);
2607 SAVEPPTR(PL_last_uni);
3280af22
NIS
2608 SAVEPPTR(PL_linestart);
2609 SAVESPTR(PL_linestr);
8edd5f42
RGS
2610 SAVEGENERICPV(PL_lex_brackstack);
2611 SAVEGENERICPV(PL_lex_casestack);
78a635de 2612 SAVEGENERICPV(PL_parser->lex_shared);
3a54fd60 2613 SAVEBOOL(PL_parser->lex_re_reparsing);
3280af22 2614
99bd9d90 2615 /* The here-doc parser needs to be able to peek into outer lexing
60f40a38
FC
2616 scopes to find the body of the here-doc. So we put PL_linestr and
2617 PL_bufptr into lex_shared, to ‘share’ those values.
99bd9d90 2618 */
60f40a38
FC
2619 PL_parser->lex_shared->ls_linestr = PL_linestr;
2620 PL_parser->lex_shared->ls_bufptr = PL_bufptr;
99bd9d90 2621
3280af22 2622 PL_linestr = PL_lex_stuff;
7cc34111 2623 PL_lex_repl = PL_sublex_info.repl;
a0714e2c 2624 PL_lex_stuff = NULL;
7cc34111 2625 PL_sublex_info.repl = NULL;
3280af22 2626
9cbb5ea2
GS
2627 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2628 = SvPVX(PL_linestr);
3280af22 2629 PL_bufend += SvCUR(PL_linestr);
bd61b366 2630 PL_last_lop = PL_last_uni = NULL;
3280af22 2631 SAVEFREESV(PL_linestr);
4dc843bc 2632 if (PL_lex_repl) SAVEFREESV(PL_lex_repl);
3280af22
NIS
2633
2634 PL_lex_dojoin = FALSE;
b27dce25 2635 PL_lex_brackets = PL_lex_formbrack = 0;
78cdf107
Z
2636 PL_lex_allbrackets = 0;
2637 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
a02a5408
JC
2638 Newx(PL_lex_brackstack, 120, char);
2639 Newx(PL_lex_casestack, 12, char);
3280af22
NIS
2640 PL_lex_casemods = 0;
2641 *PL_lex_casestack = '\0';
2642 PL_lex_starts = 0;
2643 PL_lex_state = LEX_INTERPCONCAT;
eb160463 2644 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
78a635de
FC
2645
2646 Newxz(shared, 1, LEXSHARED);
2647 shared->ls_prev = PL_parser->lex_shared;
2648 PL_parser->lex_shared = shared;
3280af22
NIS
2649
2650 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
bb16bae8 2651 if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS;
3280af22
NIS
2652 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2653 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 2654 else
5f66b61c 2655 PL_lex_inpat = NULL;
79072805 2656
3a54fd60
DM
2657 PL_parser->lex_re_reparsing = cBOOL(PL_in_eval & EVAL_RE_REPARSING);
2658 PL_in_eval &= ~EVAL_RE_REPARSING;
2659
55497cff 2660 return '(';
79072805
LW
2661}
2662
ffb4593c
NT
2663/*
2664 * S_sublex_done
2665 * Restores lexer state after a S_sublex_push.
2666 */
2667
76e3520e 2668STATIC I32
cea2e8a9 2669S_sublex_done(pTHX)
79072805 2670{
27da23d5 2671 dVAR;
3280af22 2672 if (!PL_lex_starts++) {
396482e1 2673 SV * const sv = newSVpvs("");
9aa983d2
JH
2674 if (SvUTF8(PL_linestr))
2675 SvUTF8_on(sv);
3280af22 2676 PL_expect = XOPERATOR;
6154021b 2677 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
2678 return THING;
2679 }
2680
3280af22
NIS
2681 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2682 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 2683 return yylex();
79072805
LW
2684 }
2685
ffb4593c 2686 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
bb16bae8 2687 assert(PL_lex_inwhat != OP_TRANSR);
3280af22
NIS
2688 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2689 PL_linestr = PL_lex_repl;
2690 PL_lex_inpat = 0;
2691 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2692 PL_bufend += SvCUR(PL_linestr);
bd61b366 2693 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2694 PL_lex_dojoin = FALSE;
2695 PL_lex_brackets = 0;
78cdf107
Z
2696 PL_lex_allbrackets = 0;
2697 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
3280af22
NIS
2698 PL_lex_casemods = 0;
2699 *PL_lex_casestack = '\0';
2700 PL_lex_starts = 0;
25da4f38 2701 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
2702 PL_lex_state = LEX_INTERPNORMAL;
2703 PL_lex_starts++;
e9fa98b2
HS
2704 /* we don't clear PL_lex_repl here, so that we can check later
2705 whether this is an evalled subst; that means we rely on the
2706 logic to ensure sublex_done() is called again only via the
2707 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 2708 }
e9fa98b2 2709 else {
3280af22 2710 PL_lex_state = LEX_INTERPCONCAT;
a0714e2c 2711 PL_lex_repl = NULL;
e9fa98b2 2712 }
79072805 2713 return ',';
ffed7fef
LW
2714 }
2715 else {
5db06880
NC
2716#ifdef PERL_MAD
2717 if (PL_madskills) {
cd81e915
NC
2718 if (PL_thiswhite) {
2719 if (!PL_endwhite)
6b29d1f5 2720 PL_endwhite = newSVpvs("");
cd81e915
NC
2721 sv_catsv(PL_endwhite, PL_thiswhite);
2722 PL_thiswhite = 0;
2723 }
2724 if (PL_thistoken)
76f68e9b 2725 sv_setpvs(PL_thistoken,"");
5db06880 2726 else
cd81e915 2727 PL_realtokenstart = -1;
5db06880
NC
2728 }
2729#endif
f46d017c 2730 LEAVE;
3280af22
NIS
2731 PL_bufend = SvPVX(PL_linestr);
2732 PL_bufend += SvCUR(PL_linestr);
2733 PL_expect = XOPERATOR;
09bef843 2734 PL_sublex_info.sub_inwhat = 0;
79072805 2735 return ')';
ffed7fef
LW
2736 }
2737}
2738
6f613c73
KW
2739PERL_STATIC_INLINE SV*
2740S_get_and_check_backslash_N_name(pTHX_ const char* s, const char* const e)
2741{
140b12ad
KW
2742 /* <s> points to first character of interior of \N{}, <e> to one beyond the
2743 * interior, hence to the "}". Finds what the name resolves to, returning
2744 * an SV* containing it; NULL if no valid one found */
2745
dd2b1b72 2746 SV* res = newSVpvn_flags(s, e - s, UTF ? SVf_UTF8 : 0);
6f613c73 2747
0c415a79
KW
2748 HV * table;
2749 SV **cvp;
2750 SV *cv;
2751 SV *rv;
2752 HV *stash;
2753 const U8* first_bad_char_loc;
2754 const char* backslash_ptr = s - 3; /* Points to the <\> of \N{... */
2755
6f613c73
KW
2756 PERL_ARGS_ASSERT_GET_AND_CHECK_BACKSLASH_N_NAME;
2757
107160e2
KW
2758 if (UTF && ! is_utf8_string_loc((U8 *) backslash_ptr,
2759 e - backslash_ptr,
2760 &first_bad_char_loc))
2761 {
2762 /* If warnings are on, this will print a more detailed analysis of what
2763 * is wrong than the error message below */
c80e42f3 2764 utf8n_to_uvchr(first_bad_char_loc,
107160e2
KW
2765 e - ((char *) first_bad_char_loc),
2766 NULL, 0);
2767
2768 /* We deliberately don't try to print the malformed character, which
2769 * might not print very well; it also may be just the first of many
2770 * malformations, so don't print what comes after it */
2771 yyerror(Perl_form(aTHX_
2772 "Malformed UTF-8 character immediately after '%.*s'",
2773 (int) (first_bad_char_loc - (U8 *) backslash_ptr), backslash_ptr));
2774 return NULL;
2775 }
2776
2777 res = new_constant( NULL, 0, "charnames", res, NULL, backslash_ptr,
2778 /* include the <}> */
2779 e - backslash_ptr + 1);
6f613c73 2780 if (! SvPOK(res)) {
b6407c49 2781 SvREFCNT_dec_NN(res);
6f613c73
KW
2782 return NULL;
2783 }
2784
0c415a79
KW
2785 /* See if the charnames handler is the Perl core's, and if so, we can skip
2786 * the validation needed for a user-supplied one, as Perl's does its own
2787 * validation. */
2788 table = GvHV(PL_hintgv); /* ^H */
2789 cvp = hv_fetchs(table, "charnames", FALSE);
67a057d6
FC
2790 if (cvp && (cv = *cvp) && SvROK(cv) && ((rv = SvRV(cv)) != NULL)
2791 && SvTYPE(rv) == SVt_PVCV && ((stash = CvSTASH(rv)) != NULL))
0c415a79
KW
2792 {
2793 const char * const name = HvNAME(stash);
2794 if strEQ(name, "_charnames") {
2795 return res;
2796 }
2797 }
2798
bde9e88d
KW
2799 /* Here, it isn't Perl's charname handler. We can't rely on a
2800 * user-supplied handler to validate the input name. For non-ut8 input,
2801 * look to see that the first character is legal. Then loop through the
2802 * rest checking that each is a continuation */
6f613c73 2803
b6ba1137
KW
2804 /* This code needs to be sync'ed with a regex in _charnames.pm which does
2805 * the same thing */
2806
b6ba1137 2807 if (! UTF) {
bde9e88d 2808 if (! isALPHAU(*s)) {
b6ba1137
KW
2809 goto bad_charname;
2810 }
bde9e88d
KW
2811 s++;
2812 while (s < e) {
2813 if (! isCHARNAME_CONT(*s)) {
b6ba1137
KW
2814 goto bad_charname;
2815 }
3036c853 2816 if (*s == ' ' && *(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2817 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2818 "A sequence of multiple spaces in a charnames "
2819 "alias definition is deprecated");
bd299e29 2820 }
bde9e88d 2821 s++;
b6ba1137 2822 }
3036c853 2823 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2824 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2825 "Trailing white-space in a charnames alias "
2826 "definition is deprecated");
bd299e29 2827 }
b6ba1137
KW
2828 }
2829 else {
bde9e88d
KW
2830 /* Similarly for utf8. For invariants can check directly; for other
2831 * Latin1, can calculate their code point and check; otherwise use a
2832 * swash */
2833 if (UTF8_IS_INVARIANT(*s)) {
2834 if (! isALPHAU(*s)) {
140b12ad
KW
2835 goto bad_charname;
2836 }
bde9e88d
KW
2837 s++;
2838 } else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
94bb8c36 2839 if (! isALPHAU(TWO_BYTE_UTF8_TO_NATIVE(*s, *(s+1)))) {
b6ba1137 2840 goto bad_charname;
6f613c73 2841 }
bde9e88d 2842 s += 2;
6f613c73 2843 }
bde9e88d
KW
2844 else {
2845 if (! PL_utf8_charname_begin) {
2846 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2847 PL_utf8_charname_begin = _core_swash_init("utf8",
2848 "_Perl_Charname_Begin",
2849 &PL_sv_undef,
2850 1, 0, NULL, &flags);
2851 }
2852 if (! swash_fetch(PL_utf8_charname_begin, (U8 *) s, TRUE)) {
2853 goto bad_charname;
2854 }
2855 s += UTF8SKIP(s);
2856 }
2857
2858 while (s < e) {
2859 if (UTF8_IS_INVARIANT(*s)) {
2860 if (! isCHARNAME_CONT(*s)) {
2861 goto bad_charname;
2862 }
3036c853
FC
2863 if (*s == ' ' && *(s-1) == ' '
2864 && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2865 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2866 "A sequence of multiple spaces in a charnam"
2867 "es alias definition is deprecated");
bd299e29 2868 }
bde9e88d
KW
2869 s++;
2870 }
2871 else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
94bb8c36 2872 if (! isCHARNAME_CONT(TWO_BYTE_UTF8_TO_NATIVE(*s, *(s+1))))
bde9e88d
KW
2873 {
2874 goto bad_charname;
2875 }
2876 s += 2;
2877 }
2878 else {
2879 if (! PL_utf8_charname_continue) {
2880 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2881 PL_utf8_charname_continue = _core_swash_init("utf8",
2882 "_Perl_Charname_Continue",
2883 &PL_sv_undef,
2884 1, 0, NULL, &flags);
2885 }
2886 if (! swash_fetch(PL_utf8_charname_continue, (U8 *) s, TRUE)) {
2887 goto bad_charname;
2888 }
2889 s += UTF8SKIP(s);
6f613c73
KW
2890 }
2891 }
3036c853 2892 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2893 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2894 "Trailing white-space in a charnames alias "
2895 "definition is deprecated");
bd299e29 2896 }
6f613c73
KW
2897 }
2898
94ca1619 2899 if (SvUTF8(res)) { /* Don't accept malformed input */
bde9e88d
KW
2900 const U8* first_bad_char_loc;
2901 STRLEN len;
2902 const char* const str = SvPV_const(res, len);
2903 if (! is_utf8_string_loc((U8 *) str, len, &first_bad_char_loc)) {
2904 /* If warnings are on, this will print a more detailed analysis of
2905 * what is wrong than the error message below */
c80e42f3 2906 utf8n_to_uvchr(first_bad_char_loc,
bde9e88d
KW
2907 (char *) first_bad_char_loc - str,
2908 NULL, 0);
2909
2910 /* We deliberately don't try to print the malformed character,
2911 * which might not print very well; it also may be just the first
2912 * of many malformations, so don't print what comes after it */
2913 yyerror_pv(
2914 Perl_form(aTHX_
2915 "Malformed UTF-8 returned by %.*s immediately after '%.*s'",
2916 (int) (e - backslash_ptr + 1), backslash_ptr,
2917 (int) ((char *) first_bad_char_loc - str), str
2918 ),
2919 SVf_UTF8);
2920 return NULL;
2921 }
2922 }
140b12ad 2923
bde9e88d 2924 return res;
140b12ad 2925
bde9e88d
KW
2926 bad_charname: {
2927 int bad_char_size = ((UTF) ? UTF8SKIP(s) : 1);
2928
2929 /* The final %.*s makes sure that should the trailing NUL be missing
2930 * that this print won't run off the end of the string */
2931 yyerror_pv(
2932 Perl_form(aTHX_
2933 "Invalid character in \\N{...}; marked by <-- HERE in %.*s<-- HERE %.*s",
2934 (int)(s - backslash_ptr + bad_char_size), backslash_ptr,
2935 (int)(e - s + bad_char_size), s + bad_char_size
2936 ),
2937 UTF ? SVf_UTF8 : 0);
2938 return NULL;
2939 }
6f613c73
KW
2940}
2941
02aa26ce
NT
2942/*
2943 scan_const
2944
9da1dd8f
DM
2945 Extracts the next constant part of a pattern, double-quoted string,
2946 or transliteration. This is terrifying code.
2947
2948 For example, in parsing the double-quoted string "ab\x63$d", it would
2949 stop at the '$' and return an OP_CONST containing 'abc'.
02aa26ce 2950
94def140 2951 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
3280af22 2952 processing a pattern (PL_lex_inpat is true), a transliteration
94def140 2953 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
02aa26ce 2954
94def140
ST
2955 Returns a pointer to the character scanned up to. If this is
2956 advanced from the start pointer supplied (i.e. if anything was
9da1dd8f 2957 successfully parsed), will leave an OP_CONST for the substring scanned
6154021b 2958 in pl_yylval. Caller must intuit reason for not parsing further
9b599b2a
GS
2959 by looking at the next characters herself.
2960
02aa26ce 2961 In patterns:
9da1dd8f 2962 expand:
537124e4
KW
2963 \N{FOO} => \N{U+hex_for_character_FOO}
2964 (if FOO expands to multiple characters, expands to \N{U+xx.XX.yy ...})
9da1dd8f
DM
2965
2966 pass through:
2967 all other \-char, including \N and \N{ apart from \N{ABC}
2968
2969 stops on:
2970 @ and $ where it appears to be a var, but not for $ as tail anchor
2971 \l \L \u \U \Q \E
2972 (?{ or (??{
2973
02aa26ce
NT
2974
2975 In transliterations:
2976 characters are VERY literal, except for - not at the start or end
94def140
ST
2977 of the string, which indicates a range. If the range is in bytes,
2978 scan_const expands the range to the full set of intermediate
2979 characters. If the range is in utf8, the hyphen is replaced with
2980 a certain range mark which will be handled by pmtrans() in op.c.
02aa26ce
NT
2981
2982 In double-quoted strings:
2983 backslashes:
2984 double-quoted style: \r and \n
ff3f963a 2985 constants: \x31, etc.
94def140 2986 deprecated backrefs: \1 (in substitution replacements)
02aa26ce
NT
2987 case and quoting: \U \Q \E
2988 stops on @ and $
2989
2990 scan_const does *not* construct ops to handle interpolated strings.
2991 It stops processing as soon as it finds an embedded $ or @ variable
2992 and leaves it to the caller to work out what's going on.
2993
94def140
ST
2994 embedded arrays (whether in pattern or not) could be:
2995 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
2996
2997 $ in double-quoted strings must be the symbol of an embedded scalar.
02aa26ce
NT
2998
2999 $ in pattern could be $foo or could be tail anchor. Assumption:
3000 it's a tail anchor if $ is the last thing in the string, or if it's
94def140 3001 followed by one of "()| \r\n\t"
02aa26ce 3002
9da1dd8f 3003 \1 (backreferences) are turned into $1 in substitutions
02aa26ce
NT
3004
3005 The structure of the code is
3006 while (there's a character to process) {
94def140
ST
3007 handle transliteration ranges
3008 skip regexp comments /(?#comment)/ and codes /(?{code})/
3009 skip #-initiated comments in //x patterns
3010 check for embedded arrays
02aa26ce
NT
3011 check for embedded scalars
3012 if (backslash) {
94def140 3013 deprecate \1 in substitution replacements
02aa26ce
NT
3014 handle string-changing backslashes \l \U \Q \E, etc.
3015 switch (what was escaped) {
94def140 3016 handle \- in a transliteration (becomes a literal -)
ff3f963a 3017 if a pattern and not \N{, go treat as regular character
94def140
ST
3018 handle \132 (octal characters)
3019 handle \x15 and \x{1234} (hex characters)
ff3f963a 3020 handle \N{name} (named characters, also \N{3,5} in a pattern)
94def140
ST
3021 handle \cV (control characters)
3022 handle printf-style backslashes (\f, \r, \n, etc)
02aa26ce 3023 } (end switch)
77a135fe 3024 continue
02aa26ce 3025 } (end if backslash)
77a135fe 3026 handle regular character
02aa26ce 3027 } (end while character to read)
4e553d73 3028
02aa26ce
NT
3029*/
3030
76e3520e 3031STATIC char *
cea2e8a9 3032S_scan_const(pTHX_ char *start)
79072805 3033{
97aff369 3034 dVAR;
eb578fdb 3035 char *send = PL_bufend; /* end of the constant */
77a135fe
KW
3036 SV *sv = newSV(send - start); /* sv for the constant. See
3037 note below on sizing. */
eb578fdb
KW
3038 char *s = start; /* start of the constant */
3039 char *d = SvPVX(sv); /* destination for copies */
02aa26ce 3040 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 3041 bool didrange = FALSE; /* did we just finish a range? */
2866decb 3042 bool in_charclass = FALSE; /* within /[...]/ */
b953e60c
KW
3043 bool has_utf8 = FALSE; /* Output constant is UTF8 */
3044 bool this_utf8 = cBOOL(UTF); /* Is the source string assumed
77a135fe
KW
3045 to be UTF8? But, this can
3046 show as true when the source
3047 isn't utf8, as for example
3048 when it is entirely composed
3049 of hex constants */
6f613c73 3050 SV *res; /* result from charnames */
77a135fe
KW
3051
3052 /* Note on sizing: The scanned constant is placed into sv, which is
3053 * initialized by newSV() assuming one byte of output for every byte of
3054 * input. This routine expects newSV() to allocate an extra byte for a
3055 * trailing NUL, which this routine will append if it gets to the end of
3056 * the input. There may be more bytes of input than output (eg., \N{LATIN
3057 * CAPITAL LETTER A}), or more output than input if the constant ends up
3058 * recoded to utf8, but each time a construct is found that might increase
3059 * the needed size, SvGROW() is called. Its size parameter each time is
3060 * based on the best guess estimate at the time, namely the length used so
3061 * far, plus the length the current construct will occupy, plus room for
3062 * the trailing NUL, plus one byte for every input byte still unscanned */
3063
c3320c2a
KW
3064 UV uv = UV_MAX; /* Initialize to weird value to try to catch any uses
3065 before set */
4c3a8340
ST
3066#ifdef EBCDIC
3067 UV literal_endpoint = 0;
e294cc5d 3068 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
4c3a8340 3069#endif
012bcf8d 3070
7918f24d
NC
3071 PERL_ARGS_ASSERT_SCAN_CONST;
3072
bb16bae8 3073 assert(PL_lex_inwhat != OP_TRANSR);
2b9d42f0
NIS
3074 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
3075 /* If we are doing a trans and we know we want UTF8 set expectation */
3076 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
3077 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3078 }
3079
b899e89d
FC
3080 /* Protect sv from errors and fatal warnings. */
3081 ENTER_with_name("scan_const");
3082 SAVEFREESV(sv);
2b9d42f0 3083
79072805 3084 while (s < send || dorange) {
ff3f963a 3085
02aa26ce 3086 /* get transliterations out of the way (they're most literal) */
3280af22 3087 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 3088 /* expand a range A-Z to the full set of characters. AIE! */
79072805 3089 if (dorange) {
1ba5c669
JH
3090 I32 i; /* current expanded character */
3091 I32 min; /* first character in range */
3092 I32 max; /* last character in range */
02aa26ce 3093
e294cc5d
JH
3094#ifdef EBCDIC
3095 UV uvmax = 0;
3096#endif
3097
3098 if (has_utf8
3099#ifdef EBCDIC
3100 && !native_range
3101#endif
1953db30 3102 ) {
9d4ba2ae 3103 char * const c = (char*)utf8_hop((U8*)d, -1);
8973db79
JH
3104 char *e = d++;
3105 while (e-- > c)
3106 *(e + 1) = *e;
e7214ce8 3107 *c = (char) ILLEGAL_UTF8_BYTE;
8973db79
JH
3108 /* mark the range as done, and continue */
3109 dorange = FALSE;
3110 didrange = TRUE;
3111 continue;
3112 }
2b9d42f0 3113
95a20fc0 3114 i = d - SvPVX_const(sv); /* remember current offset */
e294cc5d
JH
3115#ifdef EBCDIC
3116 SvGROW(sv,
3117 SvLEN(sv) + (has_utf8 ?
3118 (512 - UTF_CONTINUATION_MARK +
3119 UNISKIP(0x100))
3120 : 256));
3121 /* How many two-byte within 0..255: 128 in UTF-8,
3122 * 96 in UTF-8-mod. */
3123#else
9cbb5ea2 3124 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
e294cc5d 3125#endif
9cbb5ea2 3126 d = SvPVX(sv) + i; /* refresh d after realloc */
e294cc5d
JH
3127#ifdef EBCDIC
3128 if (has_utf8) {
3129 int j;
3130 for (j = 0; j <= 1; j++) {
3131 char * const c = (char*)utf8_hop((U8*)d, -1);
3132 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
3133 if (j)
3134 min = (U8)uv;
3135 else if (uv < 256)
3136 max = (U8)uv;
3137 else {
3138 max = (U8)0xff; /* only to \xff */
3139 uvmax = uv; /* \x{100} to uvmax */
3140 }
3141 d = c; /* eat endpoint chars */
3142 }
3143 }
3144 else {
3145#endif
3146 d -= 2; /* eat the first char and the - */
3147 min = (U8)*d; /* first char in range */
3148 max = (U8)d[1]; /* last char in range */
3149#ifdef EBCDIC
3150 }
3151#endif
8ada0baa 3152
c2e66d9e 3153 if (min > max) {
01ec43d0 3154 Perl_croak(aTHX_
d1573ac7 3155 "Invalid range \"%c-%c\" in transliteration operator",
1ba5c669 3156 (char)min, (char)max);
c2e66d9e
GS
3157 }
3158
c7f1f016 3159#ifdef EBCDIC
4c3a8340 3160 if (literal_endpoint == 2 &&
c286a9a6
KW
3161 ((isLOWER_A(min) && isLOWER_A(max)) ||
3162 (isUPPER_A(min) && isUPPER_A(max)))) {
3163 if (isLOWER_A(min)) {
8ada0baa 3164 for (i = min; i <= max; i++)
c286a9a6 3165 if (isLOWER_A(i))
4d73d076 3166 *d++ = i;
8ada0baa
JH
3167 } else {
3168 for (i = min; i <= max; i++)
c286a9a6 3169 if (isUPPER_A(i))
4d73d076 3170 *d++ = i;
8ada0baa
JH
3171 }
3172 }
3173 else
3174#endif
3175 for (i = min; i <= max; i++)
e294cc5d
JH
3176#ifdef EBCDIC
3177 if (has_utf8) {
55d09dc8 3178 append_utf8_from_native_byte(i, &d);
e294cc5d
JH
3179 }
3180 else
3181#endif
3182 *d++ = (char)i;
3183
3184#ifdef EBCDIC
3185 if (uvmax) {
3186 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
3187 if (uvmax > 0x101)
e7214ce8 3188 *d++ = (char) ILLEGAL_UTF8_BYTE;
e294cc5d
JH
3189 if (uvmax > 0x100)
3190 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
3191 }
3192#endif
02aa26ce
NT
3193
3194 /* mark the range as done, and continue */
79072805 3195 dorange = FALSE;
01ec43d0 3196 didrange = TRUE;
4c3a8340
ST
3197#ifdef EBCDIC
3198 literal_endpoint = 0;
3199#endif
79072805 3200 continue;
4e553d73 3201 }
02aa26ce
NT
3202
3203 /* range begins (ignore - as first or last char) */
79072805 3204 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 3205 if (didrange) {
1fafa243 3206 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 3207 }
e294cc5d
JH
3208 if (has_utf8
3209#ifdef EBCDIC
3210 && !native_range
3211#endif
3212 ) {
e7214ce8 3213 *d++ = (char) ILLEGAL_UTF8_BYTE; /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
3214 s++;
3215 continue;
3216 }
79072805
LW
3217 dorange = TRUE;
3218 s++;
01ec43d0
GS
3219 }
3220 else {
3221 didrange = FALSE;
4c3a8340
ST
3222#ifdef EBCDIC
3223 literal_endpoint = 0;
e294cc5d 3224 native_range = TRUE;
4c3a8340 3225#endif
01ec43d0 3226 }
79072805 3227 }
02aa26ce
NT
3228
3229 /* if we get here, we're not doing a transliteration */
3230
e4a2df84
DM
3231 else if (*s == '[' && PL_lex_inpat && !in_charclass) {
3232 char *s1 = s-1;
3233 int esc = 0;
3234 while (s1 >= start && *s1-- == '\\')
3235 esc = !esc;
3236 if (!esc)
3237 in_charclass = TRUE;
3238 }
2866decb 3239
e4a2df84
DM
3240 else if (*s == ']' && PL_lex_inpat && in_charclass) {
3241 char *s1 = s-1;
3242 int esc = 0;
3243 while (s1 >= start && *s1-- == '\\')
3244 esc = !esc;
3245 if (!esc)
3246 in_charclass = FALSE;
3247 }
2866decb 3248
9da1dd8f
DM
3249 /* skip for regexp comments /(?#comment)/, except for the last
3250 * char, which will be done separately.
3251 * Stop on (?{..}) and friends */
3252
c30fc27b 3253 else if (*s == '(' && PL_lex_inpat && s[1] == '?' && !in_charclass) {
cc6b7395 3254 if (s[2] == '#') {
e994fd66 3255 while (s+1 < send && *s != ')')
5ff03569 3256 *d++ = *s++;
155aba94 3257 }
c30fc27b 3258 else if (!PL_lex_casemods &&
d3cec5e5
DM
3259 ( s[2] == '{' /* This should match regcomp.c */
3260 || (s[2] == '?' && s[3] == '{')))
155aba94 3261 {
9da1dd8f 3262 break;
cc6b7395 3263 }
748a9306 3264 }
02aa26ce
NT
3265
3266 /* likewise skip #-initiated comments in //x patterns */
c30fc27b 3267 else if (*s == '#' && PL_lex_inpat && !in_charclass &&
73134a2e 3268 ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) {
748a9306 3269 while (s+1 < send && *s != '\n')
5ff03569 3270 *d++ = *s++;
748a9306 3271 }
02aa26ce 3272
9da1dd8f
DM
3273 /* no further processing of single-quoted regex */
3274 else if (PL_lex_inpat && SvIVX(PL_linestr) == '\'')
3275 goto default_action;
3276
5d1d4326 3277 /* check for embedded arrays
da6eedaa 3278 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
5d1d4326 3279 */
1749ea0d 3280 else if (*s == '@' && s[1]) {
8a2bca12 3281 if (isWORDCHAR_lazy_if(s+1,UTF))
1749ea0d
ST
3282 break;
3283 if (strchr(":'{$", s[1]))
3284 break;
3285 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
3286 break; /* in regexp, neither @+ nor @- are interpolated */
3287 }
02aa26ce
NT
3288
3289 /* check for embedded scalars. only stop if we're sure it's a
3290 variable.
3291 */
79072805 3292 else if (*s == '$') {
3280af22 3293 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 3294 break;
77772344 3295 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
a2a5de95
NC
3296 if (s[1] == '\\') {
3297 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3298 "Possible unintended interpolation of $\\ in regex");
77772344 3299 }
79072805 3300 break; /* in regexp, $ might be tail anchor */
77772344 3301 }
79072805 3302 }
02aa26ce 3303
2b9d42f0
NIS
3304 /* End of else if chain - OP_TRANS rejoin rest */
3305
02aa26ce 3306 /* backslashes */
79072805 3307 if (*s == '\\' && s+1 < send) {
ff3f963a
KW
3308 char* e; /* Can be used for ending '}', etc. */
3309
79072805 3310 s++;
02aa26ce 3311
7d0fc23c
KW
3312 /* warn on \1 - \9 in substitution replacements, but note that \11
3313 * is an octal; and \19 is \1 followed by '9' */
3280af22 3314 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 3315 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 3316 {
a2a5de95 3317 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
79072805
LW
3318 *--s = '$';
3319 break;
3320 }
02aa26ce
NT
3321
3322 /* string-change backslash escapes */
838f2281 3323 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQF", *s)) {
79072805
LW
3324 --s;
3325 break;
3326 }
ff3f963a
KW
3327 /* In a pattern, process \N, but skip any other backslash escapes.
3328 * This is because we don't want to translate an escape sequence
3329 * into a meta symbol and have the regex compiler use the meta
3330 * symbol meaning, e.g. \x{2E} would be confused with a dot. But
3331 * in spite of this, we do have to process \N here while the proper
3332 * charnames handler is in scope. See bugs #56444 and #62056.
3333 * There is a complication because \N in a pattern may also stand
3334 * for 'match a non-nl', and not mean a charname, in which case its
3335 * processing should be deferred to the regex compiler. To be a
3336 * charname it must be followed immediately by a '{', and not look
3337 * like \N followed by a curly quantifier, i.e., not something like
3338 * \N{3,}. regcurly returns a boolean indicating if it is a legal
3339 * quantifier */
3340 else if (PL_lex_inpat
3341 && (*s != 'N'
3342 || s[1] != '{'
4d68ffa0 3343 || regcurly(s + 1, FALSE)))
ff3f963a 3344 {
4d73d076 3345 *d++ = '\\';
cc74c5bd
ST
3346 goto default_action;
3347 }
02aa26ce 3348
79072805 3349 switch (*s) {
02aa26ce
NT
3350
3351 /* quoted - in transliterations */
79072805 3352 case '-':
3280af22 3353 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
3354 *d++ = *s++;
3355 continue;
3356 }
3357 /* FALL THROUGH */
3358 default:
11b8faa4 3359 {
15861f94 3360 if ((isALPHANUMERIC(*s)))
a2a5de95
NC
3361 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
3362 "Unrecognized escape \\%c passed through",
3363 *s);
11b8faa4 3364 /* default action is to copy the quoted character */
f9a63242 3365 goto default_action;
11b8faa4 3366 }
02aa26ce 3367
632403cc 3368 /* eg. \132 indicates the octal constant 0132 */
79072805
LW
3369 case '0': case '1': case '2': case '3':
3370 case '4': case '5': case '6': case '7':
ba210ebe 3371 {
5e0a247b 3372 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
53305cf1 3373 STRLEN len = 3;
06972766 3374 uv = grok_oct(s, &len, &flags, NULL);
ba210ebe 3375 s += len;
5e0a247b
KW
3376 if (len < 3 && s < send && isDIGIT(*s)
3377 && ckWARN(WARN_MISC))
3378 {
3379 Perl_warner(aTHX_ packWARN(WARN_MISC),
3380 "%s", form_short_octal_warning(s, len));
3381 }
ba210ebe 3382 }
012bcf8d 3383 goto NUM_ESCAPE_INSERT;
02aa26ce 3384
f0a2b745
KW
3385 /* eg. \o{24} indicates the octal constant \024 */
3386 case 'o':
3387 {
454155d9 3388 const char* error;
f0a2b745 3389
00ce5563 3390 bool valid = grok_bslash_o(&s, &uv, &error,
80f4111b
KW
3391 TRUE, /* Output warning */
3392 FALSE, /* Not strict */
17896a85
KW
3393 TRUE, /* Output warnings for
3394 non-portables */
80f4111b 3395 UTF);
454155d9 3396 if (! valid) {
f0a2b745
KW
3397 yyerror(error);
3398 continue;
3399 }
3400 goto NUM_ESCAPE_INSERT;
3401 }
3402
77a135fe 3403 /* eg. \x24 indicates the hex constant 0x24 */
79072805 3404 case 'x':
a0481293 3405 {
a0481293 3406 const char* error;
355860ce 3407
00ce5563 3408 bool valid = grok_bslash_x(&s, &uv, &error,
80f4111b
KW
3409 TRUE, /* Output warning */
3410 FALSE, /* Not strict */
17896a85
KW
3411 TRUE, /* Output warnings for
3412 non-portables */
80f4111b 3413 UTF);
a0481293
KW
3414 if (! valid) {
3415 yyerror(error);
355860ce 3416 continue;
ba210ebe 3417 }
012bcf8d
GS
3418 }
3419
3420 NUM_ESCAPE_INSERT:
ff3f963a
KW
3421 /* Insert oct or hex escaped character. There will always be
3422 * enough room in sv since such escapes will be longer than any
3423 * UTF-8 sequence they can end up as, except if they force us
3424 * to recode the rest of the string into utf8 */
ba7cea30 3425
06972766
KW
3426 /* Here uv is the ordinal of the next character being added */
3427 if (!NATIVE_IS_INVARIANT(uv)) {
9aa983d2 3428 if (!has_utf8 && uv > 255) {
77a135fe
KW
3429 /* Might need to recode whatever we have accumulated so
3430 * far if it contains any chars variant in utf8 or
3431 * utf-ebcdic. */
3432
3433 SvCUR_set(sv, d - SvPVX_const(sv));
3434 SvPOK_on(sv);
3435 *d = '\0';
77a135fe 3436 /* See Note on sizing above. */
7bf79863
KW
3437 sv_utf8_upgrade_flags_grow(sv,
3438 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3439 UNISKIP(uv) + (STRLEN)(send - s) + 1);
77a135fe
KW
3440 d = SvPVX(sv) + SvCUR(sv);
3441 has_utf8 = TRUE;
012bcf8d
GS
3442 }
3443
77a135fe 3444 if (has_utf8) {
c80e42f3 3445 d = (char*)uvchr_to_utf8((U8*)d, uv);
f9a63242
JH
3446 if (PL_lex_inwhat == OP_TRANS &&
3447 PL_sublex_info.sub_op) {
3448 PL_sublex_info.sub_op->op_private |=
3449 (PL_lex_repl ? OPpTRANS_FROM_UTF
3450 : OPpTRANS_TO_UTF);
f9a63242 3451 }
e294cc5d
JH
3452#ifdef EBCDIC
3453 if (uv > 255 && !dorange)
3454 native_range = FALSE;
3455#endif
012bcf8d 3456 }
a0ed51b3 3457 else {
012bcf8d 3458 *d++ = (char)uv;
a0ed51b3 3459 }
012bcf8d
GS
3460 }
3461 else {
c4d5f83a 3462 *d++ = (char) uv;
a0ed51b3 3463 }
79072805 3464 continue;
02aa26ce 3465
4a2d328f 3466 case 'N':
ff3f963a
KW
3467 /* In a non-pattern \N must be a named character, like \N{LATIN
3468 * SMALL LETTER A} or \N{U+0041}. For patterns, it also can
3469 * mean to match a non-newline. For non-patterns, named
3470 * characters are converted to their string equivalents. In
3471 * patterns, named characters are not converted to their
3472 * ultimate forms for the same reasons that other escapes
3473 * aren't. Instead, they are converted to the \N{U+...} form
3474 * to get the value from the charnames that is in effect right
3475 * now, while preserving the fact that it was a named character
3476 * so that the regex compiler knows this */
3477
ff3f963a
KW
3478 /* The structure of this section of code (besides checking for
3479 * errors and upgrading to utf8) is:
3480 * Further disambiguate between the two meanings of \N, and if
3481 * not a charname, go process it elsewhere
0a96133f
KW
3482 * If of form \N{U+...}, pass it through if a pattern;
3483 * otherwise convert to utf8
3484 * Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a
3485 * pattern; otherwise convert to utf8 */
ff3f963a
KW
3486
3487 /* Here, s points to the 'N'; the test below is guaranteed to
3488 * succeed if we are being called on a pattern as we already
3489 * know from a test above that the next character is a '{'.
3490 * On a non-pattern \N must mean 'named sequence, which
3491 * requires braces */
3492 s++;
3493 if (*s != '{') {
3494 yyerror("Missing braces on \\N{}");
3495 continue;
3496 }
3497 s++;
3498
0a96133f 3499 /* If there is no matching '}', it is an error. */
ff3f963a
KW
3500 if (! (e = strchr(s, '}'))) {
3501 if (! PL_lex_inpat) {
5777a3f7 3502 yyerror("Missing right brace on \\N{}");
0a96133f
KW
3503 } else {
3504 yyerror("Missing right brace on \\N{} or unescaped left brace after \\N.");
dbc0d4f2 3505 }
0a96133f 3506 continue;
ff3f963a 3507 }
cddc7ef4 3508
ff3f963a 3509 /* Here it looks like a named character */
cddc7ef4 3510
ff3f963a
KW
3511 if (*s == 'U' && s[1] == '+') { /* \N{U+...} */
3512 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3513 | PERL_SCAN_DISALLOW_PREFIX;
3514 STRLEN len;
3515
3516 /* For \N{U+...}, the '...' is a unicode value even on
3517 * EBCDIC machines */
3518 s += 2; /* Skip to next char after the 'U+' */
3519 len = e - s;
3520 uv = grok_hex(s, &len, &flags, NULL);
3521 if (len == 0 || len != (STRLEN)(e - s)) {
3522 yyerror("Invalid hexadecimal number in \\N{U+...}");
3523 s = e + 1;
3524 continue;
3525 }
3526
3527 if (PL_lex_inpat) {
3528
e2a7e165
KW
3529 /* On non-EBCDIC platforms, pass through to the regex
3530 * compiler unchanged. The reason we evaluated the
3531 * number above is to make sure there wasn't a syntax
3532 * error. But on EBCDIC we convert to native so
3533 * downstream code can continue to assume it's native
3534 */
ff3f963a 3535 s -= 5; /* Include the '\N{U+' */
e2a7e165
KW
3536#ifdef EBCDIC
3537 d += my_snprintf(d, e - s + 1 + 1, /* includes the }
3538 and the \0 */
3539 "\\N{U+%X}",
3540 (unsigned int) UNI_TO_NATIVE(uv));
3541#else
ff3f963a
KW
3542 Copy(s, d, e - s + 1, char); /* 1 = include the } */
3543 d += e - s + 1;
e2a7e165 3544#endif
ff3f963a
KW
3545 }
3546 else { /* Not a pattern: convert the hex to string */
3547
3548 /* If destination is not in utf8, unconditionally
3549 * recode it to be so. This is because \N{} implies
3550 * Unicode semantics, and scalars have to be in utf8
3551 * to guarantee those semantics */
3552 if (! has_utf8) {
3553 SvCUR_set(sv, d - SvPVX_const(sv));
3554 SvPOK_on(sv);
3555 *d = '\0';
3556 /* See Note on sizing above. */
3557 sv_utf8_upgrade_flags_grow(
3558 sv,
3559 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3560 UNISKIP(uv) + (STRLEN)(send - e) + 1);
3561 d = SvPVX(sv) + SvCUR(sv);
3562 has_utf8 = TRUE;
3563 }
3564
3565 /* Add the string to the output */
3566 if (UNI_IS_INVARIANT(uv)) {
3567 *d++ = (char) uv;
3568 }
3569 else d = (char*)uvuni_to_utf8((U8*)d, uv);
3570 }
3571 }
6f613c73
KW
3572 else /* Here is \N{NAME} but not \N{U+...}. */
3573 if ((res = get_and_check_backslash_N_name(s, e)))
3574 {
3575 STRLEN len;
3576 const char *str = SvPV_const(res, len);
3577 if (PL_lex_inpat) {
ff3f963a
KW
3578
3579 if (! len) { /* The name resolved to an empty string */
3580 Copy("\\N{}", d, 4, char);
3581 d += 4;
3582 }
3583 else {
3584 /* In order to not lose information for the regex
3585 * compiler, pass the result in the specially made
3586 * syntax: \N{U+c1.c2.c3...}, where c1 etc. are
3587 * the code points in hex of each character
3588 * returned by charnames */
3589
3590 const char *str_end = str + len;
3b721c4f 3591 const STRLEN off = d - SvPVX_const(sv);
94ca1619
KW
3592
3593 if (! SvUTF8(res)) {
3594 /* For the non-UTF-8 case, we can determine the
3595 * exact length needed without having to parse
3596 * through the string. Each character takes up
3597 * 2 hex digits plus either a trailing dot or
3598 * the "}" */
3599 d = off + SvGROW(sv, off
3600 + 3 * len
3601 + 6 /* For the "\N{U+", and
3602 trailing NUL */
3603 + (STRLEN)(send - e));
3604 Copy("\\N{U+", d, 5, char);
3605 d += 5;
3606 while (str < str_end) {
3607 char hex_string[4];
3608 my_snprintf(hex_string, sizeof(hex_string),
3609 "%02X.", (U8) *str);
3610 Copy(hex_string, d, 3, char);
3611 d += 3;
3612 str++;
3613 }
3614 d--; /* We will overwrite below the final
3615 dot with a right brace */
3616 }
3617 else {
1953db30
KW
3618 STRLEN char_length; /* cur char's byte length */
3619
3620 /* and the number of bytes after this is
3621 * translated into hex digits */
3622 STRLEN output_length;
3623
3624 /* 2 hex per byte; 2 chars for '\N'; 2 chars
3625 * for max('U+', '.'); and 1 for NUL */
3626 char hex_string[2 * UTF8_MAXBYTES + 5];
3627
3628 /* Get the first character of the result. */
3629 U32 uv = utf8n_to_uvuni((U8 *) str,
3630 len,
3631 &char_length,
3632 UTF8_ALLOW_ANYUV);
3633 /* Convert first code point to hex, including
3634 * the boiler plate before it. For all these,
3635 * we convert to native format so that
3636 * downstream code can continue to assume the
3637 * input is native */
3638 output_length =
3639 my_snprintf(hex_string, sizeof(hex_string),
3640 "\\N{U+%X",
3641 (unsigned int) UNI_TO_NATIVE(uv));
3642
3643 /* Make sure there is enough space to hold it */
3644 d = off + SvGROW(sv, off
3645 + output_length
3646 + (STRLEN)(send - e)
3647 + 2); /* '}' + NUL */
3648 /* And output it */
3649 Copy(hex_string, d, output_length, char);
3650 d += output_length;
3651
3652 /* For each subsequent character, append dot and
3653 * its ordinal in hex */
3654 while ((str += char_length) < str_end) {
3655 const STRLEN off = d - SvPVX_const(sv);
3656 U32 uv = utf8n_to_uvuni((U8 *) str,
3657 str_end - str,
3658 &char_length,
3659 UTF8_ALLOW_ANYUV);
3660 output_length =
3661 my_snprintf(hex_string,
3662 sizeof(hex_string),
3663 ".%X",
3664 (unsigned int) UNI_TO_NATIVE(uv));
3665
3666 d = off + SvGROW(sv, off
3667 + output_length
3668 + (STRLEN)(send - e)
3669 + 2); /* '}' + NUL */
3670 Copy(hex_string, d, output_length, char);
3671 d += output_length;
3672 }
94ca1619 3673 }
ff3f963a
KW
3674
3675 *d++ = '}'; /* Done. Add the trailing brace */
3676 }
3677 }
3678 else { /* Here, not in a pattern. Convert the name to a
3679 * string. */
3680
3681 /* If destination is not in utf8, unconditionally
3682 * recode it to be so. This is because \N{} implies
3683 * Unicode semantics, and scalars have to be in utf8
3684 * to guarantee those semantics */
3685 if (! has_utf8) {
3686 SvCUR_set(sv, d - SvPVX_const(sv));
3687 SvPOK_on(sv);
3688 *d = '\0';
3689 /* See Note on sizing above. */
3690 sv_utf8_upgrade_flags_grow(sv,
3691 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3692 len + (STRLEN)(send - s) + 1);
3693 d = SvPVX(sv) + SvCUR(sv);
3694 has_utf8 = TRUE;
3695 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */